Arm vs RISC-V: War of the platforms

The recent announcement by MicroSemi on MI-V Embedded Ecosystem that aims to accelerate the adoption of RISC-V ISA has encouraged me to write this blog. Gradually, RISC-V is building an ecosystem around the open-source ISA, and many companies, which are using Arm architecture, are including RISC-V in their portfolio. In this blog, I will try to analyse this competitive dynamics from a market structure perspective, and attempt to propose some actions for both Arm and RISC-V. Hope you can take a look at the post and share improvement areas and flaws.

Platforms

Let us take a look at market structure of the automobile industry. An automobile OEM procures raw materials such as tyres, steel, ECU, battery, etc. from various suppliers, and then assembles these parts to build the complete vehicle and sell it to end-users. This is a conventional business model or a product-based model, in which the money flows from right to left side, and the value flows in the opposite direction. The model is simple, as the transactions happen in a sequential manner.

Now, let us take cab service provider Uber. Can we have a sequential flow of transactions for Uber? No. Uber is just an aggregator that establishes connection between the driver and end-user. The value does not follow a set direction, and same is the case with money. Uber creates value for both drivers and end-users. We can say that Uber is a platform as it enables connection among different type of users, who use the service/product. There are many examples of platforms including Facebook, Twitter, LinkedIn, Google search, dating sites and more. Each of these platforms operates in a multi-sided market, mostly one side is the money-side and other side is the subsidy-side.

Adobe Acrobat can also be a platform, with readers of the pdf documents constituting the subsidy-side, as Adobe Reader is free; while content creation using Adobe Writer is charged. The idea is to get on-board plenty of users in the subsidy-side, so that the money-side is encouraged to pay for using the platform. This is an example of cross-side network effects, in which the increase of number of users on one side of the platform encourages more users to join on the other side.

The other case can be of same-side network effects. You are using Facebook because all your friends and connections are using Facebook. What is the utility to be the lone user of Facebook? If you are the only one having a phone, then whom you will call? As the number of users grows, the utility of a product/service increases, thus encouraging more users to use the product/service.

Platforms have the characteristics to develop strong network effects, and this shields competitors from entering their market. However, multiple factors have to be fulfilled to maintain strong network effects for sustainable competitive advantage.

Platforms in embedded processor market

Now that we have some basic understanding of the platforms, let us apply the concepts to embedded processor market. The various players or users in this market include vendors of ISA, CPU IPs, SoC, peripheral IPs, debug tools, operating system, EDA, boards, software applications & libraries, fabs, etc.

Among all these players, the ISA provider interfaces with multiple players in the market. On one side, it deals with the SoC manufacturers, who buy licenses for ISA, and then partner with various other players such as fabs, vendors of EDA, CPU IPs, peripheral IPs, debug tool, operating system, libraries, etc. to develop the SoC. So, the ISA is a platform, as it enables a connection among various users in the market. There is no fixed direction for flow of value and money, as many transactions happens bypassing the ISA vendor. The SoC vendors constitute the money-side that pay licensing and royalty charges for using the architecture; all other vendors, which supports the architecture and offer compatible products/services, constitutes the subsidy side. I have generalized the model for ease, as there can be exceptions wherein the ISA provider can charge few users on the subsidy-side as well.

As the number of users on the subsidy-side increases, the ecosystem around the ISA expands, thus more users on the money-side are willing to pay for using the architecture. The reverse also holds good. Thus, strong cross-side network effects kick in.

Currently, Arm architecture is a dominant platform in the embedded processor market, with an extensive ecosystem that enables strong network effects. So, Arm is de-facto architecture for many designs in the embedded market. It licenses the architecture by offering CPU IPs to SoC vendors, who partner with vendors of debug tools, EDA, operating system, etc. and fabs, to design and develop the SoC.

RISC-V, with its open-source ISA, can enable development of CPU IP that is void of licensing and royalty charges. History is replete with cases wherein open source systems have shaded the hegemony of closed and proprietary systems. Linux limited the presence of Windows to PC, while the former captured embedded, mobile and many other markets. A decade back, Nokia was at peak of smartphone market, with its Symbian operating system. However, open-source Android enabled 3^rd party developers to build apps that gave a massive boost to Android adoption. Now, Symbian is nowhere to be seen.

Should Arm be worried?

Although strong network effects shields new entrants from capturing market, some factors influence network effects.

Switching cost

A low switching cost weakens network effects. If few users move to the new platform, then more users will follow, and it will have a viral effect, leading to downfall of the existing platform. The social networking website Orkut was quite popular in India; however, as some users moved to Facebook, other users followed, and Facebook got viral adoption due to network effects. Now, Orkut is obsolete. There was no switching cost of moving from Orkut to Facebook.

Users are discouraged to move to a new platform if the switching cost associated with movement is high. Currently, many SoC vendors use Arm architecture, and their team has good learning curve on its architecture and tools. It will take time and money to switch over to RISC-V, even though the upfront cost associated with RISC-V will be less, the NRE cost and project duration may increase, as the team has to pick up knowledge of the new ISA and associated tools, which are yet to be as extensive and matured as those for Arm.

What should Arm do?

Obviously increase switching cost. Arm should explore to establish deeper relationships with SoC vendors. Instead of going for a transactional sales model, the focus should be on building long term relations. With my limited understanding, one way of doing so is to focus more on architectural licenses, as the SoC vendors will have to commit more resources and time, thus moving over to RISC-V will be difficult and expensive. Arm can explore reducing the charges or offer some favourable terms for architectural licenses.

What should RISC-V do?

RISC-V should explore how existing licensees of Arm can seamlessly switch over. One way is to offer RISC-V support for various tools, applications, operating systems, etc. that are popular among Arm’s customers.

Entry cost

High cost associated with entering a market discourages new firm to enter the market. A firm will find it difficult to enter a market due to the following reasons:

• Government regulations
• Incumbents have exclusive partnership with other firms in the value chain
• High product development cost
• Strong network effects favouring incumbents

The cost and time associated with developing a new ISA and creating an extensive ecosystem around it, is high. RISC-V started some time back as an academic project, and later ventured into professional area by steadily creating an ecosystem. RISC-V has already shown serious intentions to become a prominent player in the embedded processor market. So, we can ignore the entry cost factor, as RISC-V is way past this stage.

Multi-homing cost

Cost borne by users for associating with more than one platform constitutes the multi-homing cost. The cost can be associated with any user on the chain. Smartphone app developers, who are supporting iOS, Android and Windows, incur cost associated with building the app for each operating system. High multi-homing cost discourages users to support multiple platforms, and focus on one or few platforms that offer highest returns or maximum users. So, we see many apps on only android, as android-based smartphones are widely adopted.

Many users, who are currently using Arm architecture, are adding RISC-V in their portfolio. MicroSemi recently announced MI-V ecosystem to promote RISC-V adoption. Now, it will be offering soft-core processors for both Arm and RISC-V. Many SoC vendors are also planning to offer RISC-V based SoCs.

What should Arm do?

The money side, which includes the SoC vendors, will certainly support RISC-V, as it avoids architecture locking and may enable concession terms for licensing charges. Further, the operating cost reduces, as they don’t have to pay licensing and royalty charges for RISC-V. So, Arm should focus on the subsidy-side that includes fabs and vendors of IPs, operating system, apps, debug tools, etc. In my view, Arm can incentivize this side for promoting and supporting the architecture. Further, to increase multi-homing costs for these vendors, maybe Arm can set revenue or project target, so as to minimize the resource and time allocation for supporting RISC-V. I am not sure whether such micro-management is possible; however, a similar strategy is already validated on taxi aggregator platforms.

Some time back, Uber and Ola (Indian competitor of Uber) faced a unique challenge in India. Many drivers registered on both Uber and Ola. Drivers receive trip bookings from both the platforms, and they cherry-pick which trip they would like to take, and then decline the other one. This leads to customer frustration, as drivers refuse some trips. The solution was that both service providers started associating the drivers’ incentive to a minimum number of completed trips over a limited period of time mostly few days. With this scheme, drivers are less likely to default and refuse trips, as they will forego incentives, if the target is not met. Further, it is highly unlikely for a driver to achieve the targets on each platform within his working hours. So, this scheme indirectly increased the multi-homing cost for drivers.

Appying this strategy to the embedded market will be challenging, as the users are diverse, not uniform as in case of taxi aggregator market, wherein the strategy impacts only drivers.

What should RISC-V do?

RISC-V with its open-source ISA will definitely gather support from SoC vendors, as it offers a good-enough alternative to Arm. The existing members of RISC-V include many prominent SoC vendors including NXP, NVIDIA, MediaTek, Qualcomm, and Samsung. These members will encourage more vendors of IPs, operating system debug tools, EDA, etc. to start supporting RISC-V. So, I believe RISC-V should focus on gathering support from more SoC vendors. A week back, Andes announced the availability of 32 bit CPU IP cores based on RISC-V.

Disruptive business model

New disruptive business models have an adverse impact on network effects that are based on conventional models. Entrants can challenge the incumbents with new business models that enable more efficiency and cost optimization in the market.

With open-source ISA, RISC-V disrupted the existing market structure. The model is democratic, as SoC vendors do not have to pay licensing charges to RISC-V, so the money-side has disappeared. Value and money flow among the various users in the market, with RISC-V acting as a common base. With custom SoCs gaining prominence due to latest industry dynamics, SoC vendors can offer RISC-V-based custom SoC with less investment. Cross-side network effects will encourage more companies to support the ecosystem with supplementary product/services for enabling RISC-V-based development at low cost and fast-time to market. Read more about custom SoCs in my blog Why, How and What of Custom SoCs.

What should Arm do?

With DesignStart program and its subsequent amendments, Arm is enabling custom SoC design at lower cost and risk. The Cortex M0 and M3 CPU IPs are offered along with zero upfront license fees and a success-based royalty fee to minimize risk for commercial custom SoC development. I believe Arm should explore how to add more utility to the DesignStart program, and thus reduce the cost barriers for Arm-based custom designs.

Conclusion

In the war of platforms, both Arm and RISC-V have unique competitive advantages. RISC-V will enable collective innovation that benefits the entire community; Arm already has an extensive ecosystem with more than a decade of influence and growth. Interesting times are coming ahead; I look forward to see what strategies Arm adopts to retain its market share and how RISC-V intends to pulls the pie from Arm.

I believe there are flaws in the post, as it is based on my limited understanding of the current licensing terms of Arm. The views are completely personal; I do not have professional association with any company mentioned in the post. I believe both Arm and RISC-V have competent people, who have already thought about the likely strategies mentioned in the post. If you believe the suggestions mentioned here are useless, then I eagerly look forward to your feedback and view.

ARM vs RISC-V: Beginning of a new era

ARM vs RISC-V: A Game Theory perspective

SoC industry dynamics beyond smartphone era

Smartphones play a vital role in our life; it is sort of panacea for us. It has achieved massive growth in the last decade. With such demand of the end-product, it is obvious that the supply side has worked in overdrive. Semiconductor industry has seen massive growth due to smartphones. The number of OEMs increased to fulfil the volume demands, pulling more vendors into semiconductor industry to suffice the OEMs’ supply chain requirements.

Few OEMs made fortune in smartphone sales, with revenue exceeding billions cumulatively. Such capital inflow encouraged them to invest more money on product development, to fulfil the end-users’ paradoxical requirement of more performance with more battery life (less power consumption) at a lower price. With cumulative silicon sales going into billions, the semiconductor industry responded positively, and focused on stretching the innovation into leading process nodes and other techniques to enhance performance.

Smartphones as a forerunner

Smartphone fuelled innovation in the semiconductor industry. System on Chip (SoC) vendors were excited about the huge sales prospect and did their best to capture the market, by venturing into expensive leading process nodes, complemented with research to enhance performance without compromising on power consumption. Obviously, smartphones acted as a forerunner in terms of technology adoption. With smartphones’ shelf life of anywhere between 18-24 months, SoC vendors churn out new offerings with less lead time, as each SKU has guaranteed sales volume of millions.

General-purpose or stock SoCs achieved humongous growth in terms of sales and innovation. All these innovations have some externality, as these latest technologies spilled over to adjacent markets such as embedded and consumer electronics, and led to their growth.

Is the party over?

Currently, the smartphone growth rate is slowing down, thus the silicon demands from the OEMs will also reduce. Commoditization has also kicked in the smartphone market, so most OEMs are struggling, as end-users focus on price instead of brand.

With such industry dynamics, many chip vendors, who have just relied on smartphone as their majority revenue source, will face tough times ahead. We have already seen Imagination going off the block, after Apple pulled out of using Power VR GPUs for upcoming iPhones, maybe Apple is taking its differentiation strategy to next level. Read my related post on this: Apple, Imagination and beyond …

Has smartphones lost the slot of being the harbinger of technology?

The road ahead

With my limited understanding, smartphones as a market has lost its sheen. As the demand from the end-users will plummet, the semiconductor industry will respond accordingly. I believe that smartphones has lost the technology forerunner place to two new markets namely: IoT and AI.

Currently, the semiconductor industry is excited by the avenues of both these markets. Silicon companies are trying to associate with these technologies in some way. You can check the marketing collaterals or website of any big or small company in the chip industry, most of them will have some mention of IoT or AI, or maybe a combination of both. We are seeing a transit from a product-centric era (smartphones) to an application-centric era that encompasses the extensive usecases and applications feasible with IoT and AI. Both these technologies are placed at extreme ends on the performance scale.

IoT is not a product; it is an ecosystem of product and complementary services working in tandem to bring efficiency and optimization in any industry. Silicon sales will be largely driven by billions of low-performance IoT end-nodes or sensors that will be spread ubiquitously to collect ambient data, and then pass it on up the value chain for analysis. These end-nodes are frugal cost-sensitive devices with low power consumption.

At the other extreme of performance scale are the AI applications. These involve analysis and inference of humongous amount of data. AI can find use in diverse domains including driverless cars, ADAS, NLP, Computer Vision, Image recognition and many more. These applications need analysis of humongous amount of data for training and inference. So currently, GPU, with its parallel processing capability, are used extensively for these applications. Architectural innovation such as NVidia’s CUDA enables outsourcing of AI tasks to GPU, while the CPU can carry out system related tasks. This optimal combination of GPU and CPU on a SoC, with CUDA integrated, can be of great virtue in future, as we see AI being integrated into many applications. I believe that the industry is also getting prepared for AI. Most of the leading companies including Google, Apple, Microsoft, Intel, Facebook, Baidu, Qualcomm, along with many startups (THINCI, Groq, Nervana, etc.) are exploring the ideal way further on building AI capabilities. Hot Chips 2017 was dominated by AI and ML topics. Check a piece on EETimes here.

Edge computing, which involves processing data near to the source, is also gaining momentum with AI-powered chips, as opposed to a conventional approach that pushes data to central servers over networks, leading to transmission delays. Edge computing is useful for applications, in which real-time decisions are needed, such as driverless cars, ADAS, robotics arms, etc. AI integrated in smartphones will be one more usecase of edge computing, as this will offer real-time user experience.

Smartphones integrated with AI capabilities are work in progress. Apple is working on it. The next smartphone from Huawei will feature in-house Kirin 970 SoC that has AI processing capabilities. I am sure soon other smartphone vendors will be launching products with AI features integrated. On the other side, the power optimization innovations, which are happening on the IoT end-nodes, may be used in smartphones to extend battery life. So, smartphones, which was earlier driving technological advances in other markets, is now integrating technologies from other markets and applications.

The next technology frontiers will be driven by AI and IoT market. We will see chip industry responding to the needs of these markets by offering products that are specific to these markets. The IoT market will look for cost-sensitive and power-efficient platforms; the AI applications will be inclined for high-performance, yet power-efficient platforms. General-purpose or stock SoCs may not be ideal for meeting the requirements of IoT and AI applications. So, the industry will look beyond stock SoCs.

IoT and AI are not mutually exclusive. In certain applications, both IoT and AI can be used together. The IoT end-nodes will collect ambient data and passively pass it up to the server, which has AI capabilities, to analyse and understand the data, and then take some actions.

The inflection point

Stock SoCs have seen massive momentum due to smartphones in the last decade. However, smartphones’ sales are flattening out gradually, thus the sales volume of stock SoCs will fall southward. Simultaneously, the AI and IoT markets are picking up, and these markets need application-specific chip, not a general-purpose one. So, along with stock SoCs sales fall, we will see positive interest for custom SoCs or ASICs.

Demand for custom SoCs has also opened up avenues for small companies and startups to enter into chip design. The entry-cost barrier for custom chip design is reduced by programs such as SiFive DesignShare and ARM DesignStart. ARM DesignStart is ideal for building custom SoCs for IoT market, as the program includes low-performance Cortex M0 and M3 IPs. I didn’t see any specific markets focus for SiFive DesignShare, so maybe this program is application-agnostic. With less capital expense, custom chip design can be done. The only barrier is the high fab cost associated with production. For IoT, it is not much of a concern, as matured nodes can be used. However, AI chips will need leading process nodes for maximum performance at least power consumption. To know more about custom SoCs, take a look at my earlier post Why, How and What of Custom SoCs.

RISC-V is democratizing custom chip design with open-source ISA. SiFive, the pioneer of RISC-V, offers DesignShare program that fast tracks custom chip development, at low cost. The program relies on partners, which includes semiconductor and custom-silicon companies, to offer low-cost IPs for designing custom chips. Check my related blog – ARM vs RISC-V: A Game Theory perspective .

I do not assert that stock SoCs will lose market share completely. There are many other markets for stock SoCs. However, the sales volume of stock SoCs will definitely fall, as smartphone adoption slows down. Other markets cannot compensate for the shortfall in silicon sales.

How this is going to impact the SoC vendors, whose dominant end-market is smartphones?

Conclusion

Chip industry is capital-intensive. With large sale volume, SoC vendors can leverage economies of scale to optimize the Cost of Goods Sold (COGS). However, as the smartphone sales flatten and the industry gravitates towards application-specific products that are based on custom chips, vendors have to look for market beyond smartphones. AI and IoT are the next emerging markets. In my view, maybe not all SoC vendors can diversify their chip portfolio to cover these markets. Falling revenue will encourage the pursuit of diversification, so I believe we will see more M&A in the semiconductor industry. We are already seeing Qualcomm pursuing NXP, which is a dominant player in IoT.

Smartphone OEMs, who have backward integrated into in-house chip design, are trying to differentiate their smartphones with innovation at the chip level. Apple and Huawei have plans to launch smartphones with SoCs that supports AI. I believe Samsung and Xiaomi will also join the party soon.

Summing up, the chip indutry is in a state of flux now. As Moore’s Law is challenged by both technology and cost factors, performance enhancements are not anymore an obvious output of time and investment. We will see some real innovations happening that goes beyond the chip, and focus on the system as a whole.

I believe the post relies on many assumptions that are based on my limited understaning, so flaws are obvious. I look forward for any flaws and improvement areas.

ARM vs RISC-V: A Game Theory perspective

In any game whether it is a sport, market competition, or war, the players often start the game with their best move, watch the actions and reactions of the competing players, and then adjust their next action accordingly. So, each player takes the actions in the sequence: act, learn, and adapt. It is assumed that players are rational and they take actions within their constraints.

Market Entry

Game Theory offers excellent perspective to analyse market competition. One of the classic cases that can be modelled by Game Theory is entry by new firms into a market. The theory can be useful to explore the likely actions of both incumbents and entrants. It is expected that the players are rational, so the actions of each player can be predicted; they will both stick to respective actions that offer each of them the most optimal payoff. Under market conditions, payoff can refer to revenue, market share or both. Incumbents will be mostly inclined to gather market share aggressively, so as to shrink the addressable market for the entrant. They can pursue this objective with reduced pricing, complementary offerings, etc. Entrants do not have either market share or cash piles, so often they focus on business-model innovation or target a niche segment within the addressable market. In my view, the reaction of a dominant incumbent to entrant’s actions, justifies whether the entrant has some market potential or can poses a threat to incumbents.

RISC-V

RISC-V is a revolutionary attempt to democratize chip design. It is an open-source and royalty-free CPU ISA, which can be used to design SoC based on your requirements. You can fine-tune performance, power and price of the SoC, as per your needs. Some challenges are still there for this open-source ISA to achieve mass-adoption; however, I believe it is going in the right direction by creating an ecosystem of both hardware and software around the ISA. With RISC-V, you can design custom SoCs with very low investment.

The existing market dynamics has created a renewed interested in custom SoCs, mostly driven by frugal cost-sensitive IoT nodes and questionable Moore’s Law. So, the timing of RISC-V entry would not have been better. In case you are interested to know more about Custom SoCs, then please take a look at my post Why, How and What of Custom SoCs.

Currently there are multiple options for CPU ISA, including ARM, MIPS, Andes, Tensillica, ARC, and few more. Most of these ISAs are available with standard and custom options. Among these, ARM is the dominant player with extensive offerings that addresses from low-end to high-end applications. RISC-V, with its open-source ISA, has pursued business-model innovation that challenges the existing licensing models pursued by most other companies.

Is RISC-V really a threat for the incumbents?

ARM’s response

To answer the previous question, let us take a look at actions of the dominant incumbent ARM in the last few years.

In the last 2 years, ARM has shown renewed interest in enabling custom SoC design at low cost. ARM DesignStart program enables chip designers to build custom SoCs at low cost and fast time-to-market, by offering proven CPU IPs, along with associated peripheral IPs and subsystem. Pre-2010, ARM DesignStart program was limited to free evaluation access to Cortex-M0. In 2015, a fast-track license for Cortex-M0 was released, and the following year, an ecosystem around the program was offered with EDA partners and design partners. Recently, the program is extended to include Cortex-M3, along with zero upfront license fee and a success-based royalty fee to minimize risk for commercial custom SoC development.

Exploring the reasons

I am not aware of the exact reasons of ARM’s renewed interest on low-cost custom SoCs; however, it may be attributed to the following:

• Latest industry dynamics with a focus on customization and low-cost designs. More on this covered on my post Why, How and What of Custom SoCs
• Entry of RISC-V

It may be a combination of both as well. However, based on the timing of renewed focus on DesignStart and then followed by some improvements to the program recently, I am more inclined towards RISC-V entry as the major reason.

In the RISC-V website, I noticed that their first public event was mentioned as “RISC-V at HotChip-26” on 12^th Aug, 2014. Further, I checked some archives of leading online journals, and I found that RISC-V started getting public coverage during early second-half of 2014. And as noticed above, there is an array of updates to the DesignStart program from 2015. Read a piece on relaunch of DesignStart posted on 14^th Oct, 2015. Is it a coincidence? I don’t have a concrete answer; maybe you can share your views.

To further substantiate my view, I encourage you to take a look at the EIU The Internet of Things Business Index 2013. The report is sponsored by ARM. The report highlights the business applications of IoT-based products in various industries and revenue potential of enabling new business models with IoT sensors. One of the prime erstwhile reasons for IoT adoption was mentioned as the falling cost of underlying technology, which consists of sensors and actuators. If it was already known that low-cost is one the major drivers of IoT adoption, then why it took ARM around 2 years from 2013 to 2015, to amend the existing DesignStart program and further reduce entry barrier for designing custom SoCs that can be used to build application-specific sensors?

The post does not intend to question the strategies adopted by ARM. It is my humble attempt to apply the theoretical framework of Game Theory into a real-life business case. I strongly believe that all competitors, irrespective of their potential, should be taken seriously. Companies should aggressively devise counter-strategy to pre-empt competitors’ actions.

Irrespective of the rationale behind the updates and improvements to DesignStart, it is more important to note that the industry is benefitting from the competition. Now, start-ups and small companies have CPU ISA options to design custom SoCs based on their budget and performance requirements. Competition spurs innovation in terms of business-model and technology.

Conclusion

I strongly believe in the democratic business model of ARM, in which all the stake- holders of the fabless ecosystem makes money. In my earlier blog posts, I have shared my insights on why Intel, although having huge cash flow, failed to penetrate the smart phone SoC market. I believe the primary reason is that it was a competition between ARM’s ecosystem and Intel (with its vertically integrated business model).

Ecosystem enables network effects to take place, and it is difficult to challenge incumbents who hold important positions in the over-all value chain. Currently, ARM occupies such a position in the embedded and IoT market. For RISC-V to challenge ARM, the former must create an extensive ecosystem around its ISA. With an open-source ISA, RISC-V has opened up the value-chain further by going one step back from ARM, which earns revenue from licensing its ISA. So, RISC-V has more potential to create a symbiotic ecosystem. Read more on this competition in my blog ARM vs RISC-V: Beginning of a new era.

I believe there are lacunae in the post, as it attempts to stitch together articles from public domain and arrive at a conjecture. If you would like to share your views or you believe that the post is just a work of my vivid imagination, then I would love hear from you.

Why, How and What of Custom SoCs

Ever watched the Ted talk by Simon Sinek How great leaders inspire actions? Not yet, then I encourage you watch this 20 mins talk. This video covers the most fundamental thing that most companies fail to address: connecting with customers! Often companies focus on their products, going into details about the technical features, price, engineering innovation, etc. However, they fail to address the basic thing that is needed for a successful sale: Why they are offering the product? Answering this question bridges the gap between product and market. Revenue is an outcome, not the sole purpose of a company’s existence.

Let us take an example of a conventional sales pitch for the embedded computing platform: System on Module (SoM).

“We offer SoM that has a SoC, memory, power circuitry, Operating System, and BSPs, all integrated on a small form-factor board that offers you a platform for building your next embedded product”.

Sounds exciting? Well, it depends. However, it does not generate a great interest. Now, how about the following as a sales pitch?

“Would you like to accelerate time-to-market for your next embedded product development, while also reducing development cost and risk? Well then, we have something that may interest you”.

The latter pitch nails it, by generating curiosity and interest, yet the funny thing is that we are yet to use the term ‘SoM’, which we are supposed to sell! It starts from the customer-side, instead starting with tech jargons.

Custom SoCs are not new in the semiconductor market. However, with latest industry dynamics and enablers, there is a renewed interest in custom SoCs. As someone who is not an expert either in technology or in marketing, but maybe a jack in both 😊, this post is my humble attempt to apply Simon Sinek’s approach to custom SoCs. Hope you enjoy this post and share improvement areas on the same.

Why

Let us explore some of the industry dynamics, customer expectations, and motivations, for the push towards custom SoCs.

Questionable Moore’s Law

For more than five decades, Moore’s Law has guaranteed performance boosts, yet at lower cost and power consumption. With time, performance enhancement and power optimization was obvious. However, currently the economic equation, which is guaranteed by this law, is failing. With leading process nodes tending to reach atomic level, the designs are becoming complex, leading to long time for commercialization, thus the cost equation does not hold true. With custom SoCs, performance boosts and power efficiency is possible without compromising on cost.

Frugal and cost-sensitive IoT end-nodes

With smartphones adoption flattening out, IoT is the next wave that will drive revenue and growth of the semiconductor industry. The problem is that IoT is not a product; it is an ecosystem of products and services, each playing a specific role in the value chain. The use-cases are not dictated by the supply-side, but by the end-users’ (or the demand-side) need for optimizing cost and increasing productivity, efficiency, safety & convenience. For e.g, an industrial plant intends to reduce down-time of the system. They can put IoT end-nodes at various points, and whenever there is a deviation from the expected behavior, the relevant end-node sends alert, so the breakdown gets repaired in least possible time. The data collected from all the end-nodes can also be used to perform analytics, which helps to explore means to predict failures and increase efficiency & productivity.

Most of these end-nodes will be customer-specific and application-specific, and work in ambient conditions depending on the type of industry. Stock SoCs can be an over-kill for these niche applications, and may not fit within the budget of many start-ups and small companies working on specific IoT use-cases.

Competitive advantage

Stock SoCs are mostly a commodity as any company can use those and build their products. So, competitive advantage and differentiation are limited to mostly software. However, with custom SoCs, companies can extend the differentiation into low-level hardware as well. Some proprietary IPs, accelerators, etc. can be integrated on the SoCs to enhance performance and power efficiency.

Cost optimization

Stock SoCs are like buffet meals, in which you pay for the entire course irrespective of your appetite and food preference. As true for ala-carte option, in which you can reduce meal cost and order as per your choice, with custom SoCs, BoM cost can be reduced as you will integrate only parts that are needed for your end-product. Some reports mention that BoM cost and die size can be reduced substantially with custom SoCs.

Artificial Intelligence (AI) and Deep Learning (DL)

AI and its subset DL, Machine Learning (ML) has the potential to offer enormous possibilities. Few applications can be NLP, Computer vision, automation, etc. Currently, GPUs are mostly used to address the computation needs for AI, as the former can perform parallel processing on humongous amount of data, needed for training and inference. In future, we will see many applications and products that use AI for enhancing performance and efficiency. Many companies are venturing into custom chips that are optimized for AI. Tesla is supposed to be working on custom chips, which may have state-of-art computer vision and machine intelligence, for realizing driverless cars. TPUs from Google are optimized for AI, and some reports mention that the 2^nd gen TPUs can outperform Nvidia GPUs for AI tasks. Apple moved away from stock GPUs, and planning to build GPUs in-house, maybe to integrate more AI tasks on future iPhones.

How

In this section, I will focus on the enablers of custom SoCs. There are many proprietary ISAs that offers architectural license to build custom SoCs. However, I will cover only the two factors that are enabling penetration and adoption of custom SoCs, at low cost.

RISC-V

RISC-V ISA deserves a special attention for giving a strong boost to custom SoC market. Being an open-source ISA, RISC-V will enable designers to build custom SoCs at very low upfront cost. Designers don’t have to pay for the license and royalty. This is sort breakthrough innovation in the semiconductor industry, as it will add many new customers for custom SoCs. RISC-V is gradually building an ecosystem with IPs, debug-tools, etc.

ARM DesignStart

The ARM DesignStart enables designers to build custom SoCs at low cost. It offers free of cost access to Cortex-M0 IP, EDA tools and physical IPs. With a minimal cost, designers can license the ARM IP for production devices.

What

Finally, I would like to touch on the technical part. With my limited understanding, my view is that custom SoCs are disrupting the existing fabless semiconductor value chain. Usually, OEMs procure stock SoCs offered by the SoC design companies such as Qualcomm, Nvidia, NXP. The SoCs are designed based on some licensable CPU IPs (ARM, MIPS, etc.), then manufactured at fabs such as TSMC, Global Foundries. SoC design companies attempt to address multiple markets and applications with their stock SoCs, as this strategy offers benefit of economies of scale. Custom SoC is breaking this value chain, with OEMs bypassing the SoC design companies, and dealing directly with IP vendors, EDA vendors and fabs.

With custom SoCs, designers can integrate any proprietary IPs, analog sensors, mixed-signal parts, DSPs, accelerators, etc., depending on their application needs. This also enables designers to push many tasks to hardware, instead of doing the same in software. The benefits of pushing things to hardware are more security, reduced software memory footprint, better performance and low power consumption.

Conclusion

I strongly believe that emerging market of IoT and AI will provide a huge thrust to custom SoCs. In the last decade, stock SoCs has seen unprecedented growth with smartphones. However, in future we will see more IoT and AI applications integrated on smartphones and other embedded devices, and then maybe custom SoCs will steal the show from stock SoCs. General-purpose/ Stock SoCs will still play a dominant role; however, we will see more industry focus on application-specific parts.

It will be something like going back to past again. The industry started with ASIC, and then moved towards stock SoCs for cost advantage, and now we are seeing some motivation towards going back to custom SoCs (sort of ASIC) motivated by cost, performance and power efficiency.

Apple, Imagination and beyond…

There are multiple instances when Apple has shown the way to the industry, with breakthrough innovation in business model and engineering. Challenging the status-quo is embedded in Apple’s culture. By launching iTunes in 2003, Apple created a new business model in music industry. The model was a win-win for all – music companies, customers, artists, and Apple, of course. Then, few years later, with the launch of exciting and appealing iPhone, Apple mostly killed the hegemony of Nokia in the mobile phone industry. The popularity and massive adoption of iPhone encouraged many other companies, existing and new ones, to join the smart phone revolution.

Coming to engineering innovation that also led a race among semiconductor firms, was the introduction of 64-bit smart phone SoC. After the announcement, many other companies followed the suite.

Currently, there is one more such radical change happening. With Apple moving away from Imagination’s GPU, and planning to build its own graphics chip, the former is creating a new trend in backward integration by OEMs. Currently, there are few OEMs, which design their own SoCs in-house, including Apple, Samsung, Xioami, and Huawei; all these companies take architectural license from ARM and design the SoC. Most of these SoCs use an ARM Mali GPU. It is still unclear on Apple’s intentions of moving away from PowerVR GPU.  Is this a strategy to pull down the valuation of Imagination, making it an easy acquisition target? Or is this a move to further penetrate into backward integration by designing the GPU in-house? Whatever might be the case, Imagination may be the losing side.

Let us pick each of the options above and explore the effects of the same.

Backward integration by OEMs into chip design

PowerVR GPUs were being used in majority of Apple’s products till date. With Apple’s intention of building GPU in-house, the licensing revenues of Imagination will reduce substantially. Let us a try to take a bigger picture of this move by Apple, considering backward integration was the prime reason for this. In general, why an OEM digress away from its conventional business model of selling smart phones to end-users, and move backward into activities, which are not its core competencies?

Optimize cost

All smart phone OEMs mentioned above has high sales volume, going into hundreds of millions. With such humongous volume, it makes sense to control your recurring operational cost, which also includes licensing rents and royalty. OEMs have to trade-off between Build vs Buy. In large volume, it is pragmatic to adopt Build strategy, instead of Buy, which will cause a continuous cash outflow. With a one-time, yet high NRE cost, OEMs can design a SoC in-house. Obviously, there will be still some continuous expenses such as engineering resources salaries, manufacturing cost, etc. However, designing a SoC in-house optimizes cost in the long run, as the high NRE cost amortizes over large sales volume. Custom SoCs also reduces BoM cost as only required components are integrated on the SoC.

Competitive advantage

With custom SoCs, OEMs can differentiate their products from those of the competitors. Such differentiation is not possible by using standard off-the-shelf SoCs available in the market. OEMs are close to the end-users, so they know the end-user’s expectations better than conventional SoC design firms. SoC design firms have large overheads, so it makes a good business case for them to design standard SoCs and achieve high sales volume by targeting multiple use cases, markets and applications. Backward integration into SoC design gives flexibility to the OEMs to fine-tune their products to meet customer requirements and differentiate their products from those of competitors. In-house SoC design also enables OEMs to have their custom implementations, suited to their performance and power requirements.

With the recent fallout with Qualcomm on modem licensing revenues, would Apple attempt going into designing modem chips? With my limited knowledge, it may be far-fetched and over-ambitious for getting into modem development. Apple still continues to source from Qualcomm. With respect to in-house development of GPU, Imagination holds the view that Apple cannot design a GPU without violating Imagination’s patents. Let’s see when wewill see Apple products with in-house GPU.

Machine Learning and AI

Machine Learning and AI are now hot trending topics in the technology sector. Companies are still wondering how to monetize these technologies. We will definitely see considerable actions in these fields in the near future. There are endless possibilities of these technologies to be used for various applications such as NLP, Computer Vision, Image recognition, and many more. These applications requires massively parallel computations, which are most suited to run on GPU, rather than hogging the CPU. I believe this maybe one more reason that encouraged Apple to move away from standard GPU, and develop in-house GPU, which are optimized for machine learning and AI. In future, we will see more and more AI features integrated in smart phones. With in-house GPU, maybe Apple is preparing for the future.

Possible takeover of Imagination

This point is more specific to this case, unlike the earlier point, which broadly analyzed the backward integration strategy followed by OEMs. The move by Apple has hit the stock prices of Imagination and few analysts claim that the latter will be loss-making by 2019, after the removing the revenues from Apple. Maybe this is a deliberate move by Apple to shake up the GPU vendor, and then acquire it at a low price. Apple will get access to patents of Imagination, and may continue using PowerVR in its product portfolio.

Should ARM worry?

Assuming that Apple takes over Imagination, should ARM be worried? I strongly believe that with its huge market share, Apple has the strength to influence the semiconductor market. After acquiring Imagination, Apple gets access to MIPS ISA. Currently, ARM is almost a monopoly ISA for smart phone SoCs. However, with a huge market presence and willingness to take bold steps, Apple will be at bargaining table with ARM, and Apple may consider spending millions to make a switch from ARM to MIPS, in its products. Although MIPS ISA has an almost negligible share in smart phone SoC market, with Apple supporting this ISA, things may go against ARM. So, with takeover of Imagination, Apple gets access to PowerVR GPU and MIPS CPU. What do you think about this?

Conclusion

I believe Apple can greatly influence the dynamics of the semiconductor industry. Irrespective of which of the two options Apple takes further, I believe that many other companies in this industry will face the ripples.

I know that things are not as simple, as suggested in the post. I will be overly glad to receive loopholes and improvement areas on this post.

ARM vs RISC-V: Beginning of a new era

The emerging Internet of Things (IoT) industry is an aggregation of products and services, complementing each other to enable efficiency and cost-optimization in multiple industries. It does not have a vertically oriented value chain. It stretches horizontally across multiple industries and markets such as Industrial automation, Automotive, Medical, environmental monitoring, and many more. The use-case for these industries are quite diverse. At the front-end of the applications are end-nodes or sensors, which monitor the ambient conditions and pass on the data down the chain. These end-nodes will be scattered in billions in various industries.

Rise of Custom processors

As mentioned in my earlier post ARM vs Intel: The new war frontiers, COTS processors will not be ideal for building these end-nodes, as the latter are application-specific. Companies would be inclined to custom-processor, as it offers flexibility to assemble only required parts. These parts can include analogue sensor, DSP or proprietary IP, etc. Further, custom processors substantially reduce BoM cost and die-size, which will minimize power dissipation. It also helps companies to differentiate their product from those of competitors. The low entry cost and pervasive presence of IoT industry will encourage many start-ups and small companies to build products for niche applications. With custom processors, these companies can further optimize the cost.

Failing Moore’s Law

In addition to the proliferation of the IoT devices that will give a huge boost to the custom processors, another influencing factor is questionable existence of Moore’s Law. For more than five decades, Moore’s Law acted as a self-fulfilling prophecy. Semiconductor companies raced to make this law a truth, irrespective of whether the market needs high performance processors. There are always the innovators and early-adopters that are desperate to use products based on leading process nodes. However, the mass market takes time to switch to these new products. With high-performance, reduced power consumption, and cost reduction, Moore’s Law ensured that technology played a dominant role.

However, currently the economic equation that is guaranteed by this law, is failing. Leading process nodes are becoming complex to design, with long lead time for commercialization, thus the cost equation does not hold true. The quest for cost optimization forced the industry to look for alternatives, as shrinking nodes is no longer economically beneficial. Customization is the answer, as it can reduce the BoM cost significantly. The billions of end-nodes does not need leading process nodes, custom processors at matured nodes will be good enough.

ARM’s response

ARM is a monopoly in smartphone processor market. In the embedded and IoT space, there is no dominant architecture yet. ARM is ideally poised to fill up this vacant position, as it already has a strong presence with CPU IP offerings at diverse power, performance, and price options. With DesignStart license for Cortex-M0, ARM has enabled custom processor design at low cost with less risk. This program will be really useful for start-ups and small companies, as they get an access to proven architecture and IPs at low licensing cost, complemented along with an extensive ecosystem of IPs, software support and silicon partners that can massively reduce the time-to-market for products.

How we can further optimize cost of custom processors?

Enter RISC-V

Open Source Software (OSS) has played a vital role in democratization of the software industry. One of the most popular examples of OSS is Linux operating system. OSS enables innovation and differentiation, at a low cost of adoption. This promotes small firms, start-ups to build products based on OSS such as Linux. A large community of developers supports the software development, so there is no risk of vendor-locking or obsolescence of proprietary technology. The collective efforts of the community ensure a large ecosystem, and this benefits all the users. Linux has gained enormous presence in diverse applications such as embedded, PC, etc. Network effects can also be leveraged, as more users start using Linux, more features, and utilities are added.

RISC-V extends the open-source movement into CPU ISA.  It is an open-source ISA that is license-free and royalty-free. As RISC-V is void of any licensing, the ISA can be used for building custom processors with zero licensing cost. RISC-V is gradually building an ecosystem. In Embedded World 2017, RISC-V showcased the extensive ecosystem with FPGA solutions, security IPs, debug infrastructure, etc.

Few ARM customers have already started using RISC-V for designing custom processors. Now, SoC design companies can develop custom processors at a low cost, without the paying licensing rents. With some NRE investment, these firms can develop the SoC and get it manufactured in fabs. Thus, the price of the processor will be lesser than those based on ARM IPs. At the face value, it looks like we have found an ideal candidate that has the potential to become the dominant ISA for IoT industry. With customization & zero licensing cost, RISC-V looks like a winner.

Is it really that simple?

Are there free lunches?

Linux is a quite successful with billions of deployment in diverse products. Although, it needs considerable effort and expertize in using Linux for commercial products, the benefits weigh over the man hours. Linux offers unmatched flexibility. The enormous community provides a good ecosystem around the OS, with extensive support for peripherals, 3^rd party software, etc.

However, extending the concept of open-source to chip design is a different ball game, owing to the basic differences between software and hardware. Unlike software, which needs only time and effort to be developed, hardware involves tangible components, for which someone has to pay. Secondly, you can rework on the software any number of times after testing on the hardware, emulator, etc. Bugs can be fixed with only investment in time and effort, maybe with minimal cost. However, bugs in the hardware can be a million dollar loss! Multiple spins of a processor can shoot cost substantially. Finally, hardware design is more complex than software development.

Let us consider the case of open source RISC-V. In a SoC, CPU IP is just one part; there are many other physical IPs and peripherals needed. So, an extensive IP and EDA ecosystem is needed around the CPU IP.  You can only get the CPU IP without licensing rents; however, the surrounding ecosystem is gone cost you. IP vendors should see a viable business case to add support for RISC-V in their portfolio. Let us assume that a strong community backs the RISC-V, and it offers all the IPs and tools needed for building SoCs. However, the question remains whether companies building custom SoCs, will take the risk of using a community backed ISA? Bugs can lead to multiple tape-outs, which add huge cost. Finally, designing a SoC is complex and needs good expertize in multiple areas such as implementation, physical design, packaging, etc.

With ARM ISA, most of the issues mentioned above are alleviated. You get access to proven IPs, robust ecosystem (software, cloud services, security solutions, silicon vendors, fabs), and committed support, instead of community support offered by open-source ISA. Design complexity is reduced; however, still some expertize in SoC design is needed for building custom processors.

Who will build RISC-V based SoCs?

The idea of open-source is disruptive, as it enables a level-playing field to companies, with limited budget, to compete against big players. Although the concept of open-source ISA is revolutionary, it may not have a disruptive effect in democratizing chip design.

In my view, it is unlikely that small companies and start-ups addressing some niche application in the IoT space, will invest time, effort and money in building custom processors based on a community-backed ISA, as they have to validate whether the entire system meets their specifications. It would be a safe bet for them to use licensed ISA, as they get a proven system, complemented with a robust ecosystem. Multiple tape-outs of the SoC can add substantial cost. A matured ISA with some initial cost would be a good starting point, instead of a free fledging ISA. SoC design is not their core activity, so hiring a diverse team for chip design, may not be a pragmatic decision. ARM is used widely across the industry, so the design part can be outsourced to some small companies, specializing in ARM-based SoC design. EDA tools and fabs are costly. EDA vendors and fabs already support ARM-based IPs; they should see economic benefits for adding support for RISC-V. Until RISC-V reaches a critical mass adoption, it’s like a chicken-and-egg situation. Multi-homing adds cost for any company be it a fab, EDA vendor, design firm, or an app developer. Low volume business will attract higher rents. All these cost overheads have to be factored while building RISC-V based SoCs.

Market leaders in the SoC design will definitely develop RISC-V-based SoCs, as it increases their buying power by having some alternative to ARM. However, I believe that these companies will not be interested in engaging with low-volume customers, who needs custom processors. Owing to their large overhead, it makes business sense to sell millions of standardized SoCs.

Summing from above, in my view RISC-V, in its current state, cannot significantly disrupt the semiconductor market structure. One of the key virtues of open-source movement is minimize entry barriers into a market by offering a good enough base, in comparison to licensed entities. Although, RISC-V will offer flexibility for building custom SoCs at low cost, the ecosystem is not yet ready to accept it. The whole semiconductor industry need to work in sync to make RISC-V successful.

Conclusion

It is too early to pronounce the verdict on RISC-V. Any new idea takes time to flourish. I am sure the people at RISC-V are smart enough that they would have foreseen the issues that I have mentioned above and many of the issues would have been already mitigated internally. In my view, RISC-V should focus on one segment such as IoT end-nodes or something else, and then offer this segment a compelling and complete solution, along with a holistic ecosystem, rather than focusing on the entire IoT and embedded industry. Once they achieve mass adoption in one segment, it is easier to spread into other segments, as new users have a great case study or example to look upon.

What should ARM do better to be perceived as a leader in the embedded and IoT segments? I do not have any answer for this, as from an external perspective things look pretty well now for ARM, with a huge installed base. The upward trend will continue in future as well. Extending DesignStart license to other Cortex-M IPs would be a good option for further adoption. However, the main forte should be the strong ecosystem of OS support, cloud services, security, IPs, debug toolchain, EDA, silicon partners, etc. All these play a vital role in building a successful product based on a custom processor at low cost.

Low-cost and customization are often mutually exclusive. Any ISA addressing both these ends will play a dominant role in the IoT industry. My views are limited to my knowledge. I believe there will be many lacunae in this post, so I look forward to the improvement areas. I do not have any professional obligations toward any companies mentioned in this post. The views expressed are completely personal.

Embedded Product Development – Make vs Buy

Original Equipment Manufacturers (OEMs) face many questions before building any product. After they are convinced that there is a business potential in their new product, next comes the crucial stage of project execution. They aspire to build the product in-time, maybe before the competitors or better than the competing products, without compromising on their budget constraints. However, aspirations occasionally match with reality. Time-slips, production-failures, over-budgets, etc. are associated with most projects. In this blog, I will attempt to show how using COTS (Commercially off the shelf) platforms can help OEMs accelerate the time-to-market along with reduction in development cost and risk.

Cost, performance, PCB designs, memory, time-to-market, technical support, casing, I/O configuration, size, procurement, enclosures, flexibility, scalability, component obsolescence, compliance, certifications. Whew the list goes on! You will face many more questions while building an embedded product. With customers’ expectations for better performance, yet longer battery life, is making the product development increasingly more complex. Advances in technology are inevitable, and this forces OEMs to keep pace with technology and competitors. Complex designs on a small form-factor adds substantial design risk, which may further stretch the development time. However, using COTS platforms reduces your list of concerns substantially.

Any embedded product has mostly similar components, both for software and hardware. Hardware includes SoC, memory, power circuitry, I/Os (USB, Ethernet, VGA, WiFi, BT, etc.) integrated over a printed circuit board (PCB). The software consists of device drivers, operating system, BSPs, GUI, application layer, 3^rd party apps, communication stack, etc.

Make: Full-custom Development

OEMs are more inclined to build products from scratch, as it offers total flexibility and better control over quality, cost. However, full-custom development has many constraints.

• Boost NRE (Non-Recurring Engineering) cost: High investment in engineering resources as the team need diverse expertize in mechanical design, hardware layout, low-level firmware, application, etc. More testing and validation is needed, as the product is developed from scratch. Hardware design iterations add to substantial project cost as well.
• High BoM cost: Usually, sales volume of embedded products is low. So, OEMs cannot leverage economics of scale with low volume procurement of components, and thus Bill of Materials (BoM) cost is high. However, if the sales volume exceeds 50-60K per year, then it makes more investment sense to pursue full-custom designs.
• Long time-to-market: As the development happens from scratch, the project time increases, and thus long time-to-market. Multiple hardware design iterations compound the time-slip.
• High development risk: With scratch development of hardware and software, there is a high probability that things may go wrong at any level. This adds significant risk to the project compromising time-to-market and development cost.
• Questionable scalability: With Moore’s Law in action, the silicon components such as SoC, are getting matured in terms of performance, power-efficiency, and cost-effectiveness. However, it is difficult to scale up a full-custom platform to accommodate these advances. Upgrades to a platform based on future customers’ demands and latest technologies may need re-design.
• Questionable Product Life: Although designers pursue multiple sourcing of components, once a critical component such as SoC, RAM, and Flash, reach End of Life (EOL), a re-design will be needed to accommodate the substitute component. For industrial products, component obsolescence management is critical, as the product life is more than 10 years. So, each and every component used in platform must be available for this extended period. This adds substantial overhead for designers in terms of supply-chain management.

Buy: COTS Platforms

Let us now explore the COTS platform and their advantages. COTS platforms such as Single Board Computer (SBC) and System on Module (SoM) are available with the hardware platform and low-level software including Operating System, BSPs and Device drivers. OEMs can focus on enhancing user experience with awesome GUI, application-specific frameworks, etc., instead of engaging in generic board bring-up activities. There is no value addition in reinventing the wheel, as once an operating system is supported on a SoC, the BSPs can be reused.

Advantages over Full-custom Designs

• Lower NRE cost: As the hardware and associated low-level software are already available, the scope of project reduces. Focus will be on integration and application development. Thus, the resource cost, along the validation effort, comes down significantly.
• Lower BoM cost: COTS platform vendors leverages economies of scale in component procurement and manufacturing, with huge volume. Thus, the platforms are cheaper than sum of individual parts’ cost.
• Accelerate Time-to-Market: The project timeline becomes shorter, which accelerates time-to-market. Further, the low-level software and hardware are already matured, so bugs are limited mostly to the application layer.
• Lower Development Risk: The platforms are validated the vendors and numerous existing customers, thus the platforms are robust and matured.
• Long Product Life: Usually, vendors guarantee the platform availability over extended time period. In case, any component reaches EOL, then it is the vendors’ responsibility to ensure availability of the platform. OEMs need not have to worry about critical component obsolescence.
• Less supply chain overhead: System designers need to deal with less component vendors, as most of the critical components are available on the COTS platform.
• Access to latest technologies: Usually, market leaders of semiconductor parts such as SoC, Flash, and RAM prefer to engage with customers, who order for large volume. However, most embedded products have low sales volume. By using COTS platforms, OEMs can get access to latest technologies from leading vendors, as COTS platform vendors engage in huge volume semiconductor business.

Single Board Computer (SBC)

Single Board Computer (SBC) is used widely in embedded computing industry to build variety of products. SBCs are off-the-shelf, application-ready embedded platforms that host the processor, memory, power circuitry, and I/Os on a single printed circuit board (PCB), and comes along with associated device drivers, Operating Systems and Board Support Packages (BSPs). So, the product development becomes fairly simple. System designers can just build the application software and put the board in a nice enclosure, then the product is ready.

However, there are few constraints of using SBCs for embedded product development.

• Not scalable: In a SBC, the processing unit and I/O section are integrated over a single PCB. So, it is not possible to migrate to a latest processor with the same board. For migrating to latest technologies or meeting customers’ future expectations, new SBCs have to be used.
• Not flexible: Customizing a SBC, based on the OEMs’ requirements, is not possible as the CPU and surrounding I/O are closely coupled due to the single-board design. Usually, standard I/Os are part of the SBC. Additional peripherals can be added using interface boards; however, this may increase the size of the platform. Further, the I/O configuration is fixed, so it is challenging to build size-constrained products.

System on Module

An embedded platform can be represented as below:

The ‘Application Agnostic’ part consists of essential design commodities, including the processing & memory requirements. This part may not differ much whether the end-product is a medical device or retail PoS device, assuming the processing and memory requirements are somewhat similar.

This ‘Application Specific’ part constitutes both the hardware and software, depending on the end-product and OEM requirements. OEMs can enhance end-user experience by creating awesome UI, user application, etc.

A Computer on Module (CoM) or System on Module (SoM) is an embedded computing solution that consists of the application-agnostic hardware and software. System designers can focus on the application-specific part by using an off-the-shelf SoM, and thus accelerate time-to-market. The combination of an application-agnostic SoM and application-specific carrier board, along with display and peripherals, offers a complete platform for building any end-products. OEMs can design carrier boards as per their size and I/O requirements. The SoM can be inserted into the carrier board through some standard connector such as SODIMM or MXM.

In addition to the generic benefits of COTS platform, SoM also resolves the scalability and flexibility issues inherent to SBC.

• Platform scalability: Most vendors offer pin-compatible SoMs. This means that a carrier board can be used along with multiple SoM, without any hardware changes. Some application software change may be needed. This ensures seamless migration to latest technologies. For example, a ECG machine is launched in the market; however, after 2 years due to market demands, the OEM intends to use the latest and faster processor. Without a platform redesign, the OEM can easily migrate to the latest technology by using a SoM based on the new processor, on the existing carrier board. Thus, platforms remain future-proof. Further, product variants with different performance and price can be launched without full-scale development for each variant.
• Platform flexibility: Each application has specific requirements in terms of I/Os, size, performance, and power. The OEM can select an off-the-shelf SoM based on the performance and power needs. The carrier board can be custom-built as per the I/O and size requirements. Thus, SoM approach offers more flexibility than the SBC approach.

Conclusion

We can summarize that in the choice between ‘Make vs Buy’ for embedded product development, the ‘Buy’ option is more favorable in terms of time-to-market, development cost, obsolescence management, development risk, and scalability. Further, among the COTS platforms, SoM are better equipped to handle the demands of projects instead of SBC.

As usual, this post is constrained by my bounded rationality. Please share the improvement areas and flaws on this post.

ARM vs Intel: The new war frontiers

With Intel’s exit from smartphone processor market, the competitive zones are redefined with its rivalry with ARM. Is ARM’s domination the only reason for Intel’s exit? With no competing architecture, is ARM a monopoly in smartphone processor IP market? What are the new areas of competition between ARM and Intel? I will attempt to answer these questions in this post.

Intel’s exit

I read some news that Intel invested around $10 billion in its mobile processors endeavors.  Intel not only invested such huge capital but also digressed from its conventional business model to gather market share. For gaining traction in the low-cost smartphones, SoFIA project was started. In this project, Intel relied on partners, putting aside its vertically integrated business model. Intel partnered with 3^rd party SoC design companies for designing SoC on the Atom CPU, and these SoCs would be manufactured in the 3^rd party fabs.

It is no brainer to say that ARM is the major reason for Intel’s exit. Can there be some more reasons? In order to compensate for its late entry into smartphone market, Intel paid smartphone OEMs to use its SoCs, and focused on all segments from cost-sensitive volume market to premium segment. The text-book strategy to penetrate a market with dominant incumbent is to focus one customer segment and offer them a complete solution. Did Intel’s strategy match this? I don’t think so. Intel attempted to cater the needs of wide customer segments from high-end to low-end. The solution was not complete, as a competitive integrated modem was missing in some SoCs. I wish Intel would have followed the text-book strategy. It would have been a good experiment to test the credibility of such text-book theories in practice.

Anyways, this is a retrospective discussion for now. Intel also never got any good deal from major OEMs such as Apple, Samsung and Xiaomi. One of the reasons behind this is that all these OEMs have also backward integrated and started designing their own processors. So, x86 is not an option. These companies take architectural license from ARM, and then design the processor as per their needs, and manufacture the processor in fabs such as TSMC, Samsung, etc. The impact of OEMs’ backward integration to pure-play SoC design companies such as Qualcomm, Mediatek, merits a separate analysis. I will try to focus on this in my next post. Open-source systems boost innovation and adoption, as evident from the wide-spread adoption of entities such as Android, Raspberry Pi, Linux, etc. A closed system can never achieve the benefits of network effects, which dictates that value of an entity increases as more users adopt the entity. Intel can never achieve this, as x86 is proprietary.

Is ARM now a monopolistic power?

With the exit of Intel from smartphone market, ARM is the only company offering IPs for smartphone processors. Does this mean smartphone processor IP market is now a monopoly? Can ARM attract monopolistic rents? With my limited knowledge on microeconomics, I believe the smartphone processor IP market can be modelled as a monopoly, as only one company is active. ARM licenses its IPs to many companies that design processors based on those IPs. ARM makes revenue through licensing cost and royalty. What prevents ARM to dictate prices for smartphone processor IPs?

In my view, there are primarily 3 reasons.

• Dwindling smartphone market: Smartphone penetration has almost reached saturation in developed market. Most sales happen for add-on or replacement phone. Now, developing nations such as India and China holds the next new billion potential customers. These markets are price sensitive, so driving price to cash-on to the monopolistic power by ARM will have negative impact on sales volume. Rising input cost will result in higher end-product price

• Most SoCs are application-agnostic: ARM license processor IPs that can be used by the design companies to develop SoCs targeted for variety of applications. For example, the same SoC can be used for smartphone, gaming consoles and high-end computing applications. So, it is not possible only to rise licensing and royalty fees for smartphone processor IPs

• Smartphone as a growth driver for adjacent or emerging market: The pervasive presence of smartphone has led to growth of many adjacent or emerging market. Smartphone acts as a communication medium between humans and machines/ devices. The rising IoT era is dependent on smartphone for connecting the tiny end-nodes with humans. Wearable devices such as blood pressure monitoring, etc. need to be compact, power-efficient and low-cost. So, most of these devices are headless and use the smartphone as a medium to communicate with human, using some software app.

The same is true for embedded computing market that includes applications for home-automation, industrial automation, medical, automobile, etc. A diverse range of these devices depends on smartphone. In home automation, a remote assistance product for elderly people uses the care-taker’s smartphone to send distress signals, in case the elderly people meets with an accident.

IoT is the next emerging market that will give a great boost to electronics / semiconductor industry, so the enablers or growth drivers of this market should be cost-effective to drive mass-scale adoption. ARM supplies low-cost processor IPs such as Cortex M series that are widely used for connected devices. A rise in smartphone processor IP, usually the high-end ones such as Cortex A15, A9, A53, etc., may decelerate the growth of smartphone market, in turns will restrict the growth of the adjacent markets

New war frontiers

With Intel’s exit from smartphone processor, the areas of competition between Intel and ARM have shifted grounds. Intel will be focusing more efforts on the emerging connected and IoT market. IoT is the next tech tsunami that will drive innovation, competition and market penetration, as smartphone has done so in the last decade. The challenge is that IoT is not a product, it is an ecosystem. Apart from the software offerings such as cloud management, APIs, data analytics, etc., at the hardware level it needs sensors or end-nodes, gateways, server or data center, and connectivity devices.

At the front-end or customer-facing end of IoT applications are connected things or sensors, that tracks or monitors some ambient conditions. These end-nodes or frugal devices will be scattered in all places from light bulbs, vehicles, building, elevators, industrial plants, oil refineries, and many more. The data collected from the ambient environment will be passed on to gateways that may do some processing on the data. The data is pushed from the gateways to servers, in which advanced data analytics can be done on petabytes of data. Cloud services plays a vital role in this chain from data collection, reception and analysis. Each of the hardware involved in the chain including sensors, gateways, and servers will be having some processing unit, memory and some I/Os depending on their usage.

Sales volume will be driven by end-nodes or sensors that will be needed in billions for collecting vital information from various surroundings. To drive mass adoption, these end-nodes should be low-cost; however, due to application-specificity and customer-specificity, the sales volume for individual SKUs will be less. COTS processors may not fulfill all the requirements of power, performance, price and space of the end-nodes. The focus will be on application-specific processor that is tuned to the needs of the specific industry. Companies building IoT products would be more interested in custom processors that are tuned to their requirements and budget. However, the problem is to fulfill the conflicting objectives of customization and low-cost. Any company solving this problem will crack the IoT market.

Now coming to the competition in the end-nodes business between ARM and Intel. Intel offers Quark SoC and micro-controllers that can be used for the end-nodes or connected things. In a recent blog after its restructuring, Intel made clear that it will work towards offering a complete ecosystem for IoT: frugal connected things, servers, networking and cloud services. However, one thing that I fail to understand yet is how Intel will control the cost structure of Quark or similar processors, without making business loss? With its vertically integrated business model, it is difficult to achieve cost-optimization. Further, off-the-shelf processors may not be ideal for many IoT applications.

In past, for capturing low-cost smartphone market, Intel sidelined its vertically-integrated business model and started the SoFIA project, in which it partnered with Chinese design companies for the SoC design and then get the SoC manufactured at pure-play foundries. The idea was to cost-control that is not possible with the high-margin Intel’s business model. Shall Intel pursue a similar program for low-cost processors targeted for the connected things? I believe, it is worth a try.

However, the second issue that may erupt with the above strategy is that COTS processors may not be an ideal choice for many IoT applications. ‘One processor fit all’ may not hold good for IoT market, as the applications are diverse in terms of usecase, power, performance, and I/Os. With proprietary x86, it is not possible to build customized SoCs. Should Intel start licensing x86 architecture? This is a difficult question to answer, for me at least. However, I strongly believe in the power of network effects, collaboration and coo-petition. All these cannot be possible with closed standards. AMD has already started licensing x86 server IPs. I am sure that Intel will have to deal with more competitors in the near future. What do you think on how Intel can address this issue?

On the other hand, ARM is steadily carving an IoT ecosystem. Already being used in billions of smartphones and embedded devices, ARM may have an edge over Intel in the IoT market currently. With IP licensing model, system designers can differentiate their products from those of competitors by designing processors according to their budget, form-factor and application requirements. This offers better control to companies to target niche segments of IoT. ARM has realized that mass-scale production by large SoC vendors will not bring radical innovation in IoT space, instead value-addition will be driven by small companies that focus on niche markets.

With ARMDesignStart program, ARM is lowering the entry barrier for custom SoC design and manufacturing on Cortex-M0 IP. This program offers low-cost access to Cortex-M0 IP along with design services (from Cadence, & Mentor Graphics) and physical IPs. This will promote further penetration of ARM IPs among start-ups, design enthusiasts, etc. Designers can start their development with limited fixed cost. After evaluation, they can go for full-scale production at nominal cost. This will enable low-cost products targeted to hundreds of niche IoT applications.

Further, ARM also aspires to capture market share in server space. Presence in server space will be enable a full-scale IoT value chain: end-devices (Cortex M series), gateways (Cortex A9, A15 etc. in vehicles, medical, & many more), servers (64 bit processors) and software (mbed, Cloud services).

Conclusion

Interesting times are coming ahead. Both companies, along with many others will compete to realize the billions of connected devices. I strongly believe most innovation will happen at small companies and start-ups, rather than in large companies. The strategy should be address the needs of these thousands of small companies by offering a base that is cost-effective and scalable.

I really look forward to your views, suggestions and improvement areas. I would like to emphasize that these are completely personal views with my limited knowledge. I do not work for any organization mentioned in this post. I also encourage you to take a look a look at my earlier blogs on ARM vs Intel.

ARM vs Intel – Not a tech war

ARM vs Intel – The way further

 

Are Smart things making us smarter?

Nowadays, we don’t have to learn how to drive a vehicle well because there are systems in the car that is taking care of many things without our knowledge.  We don’t have to remember whether we have switched off the lights before leaving the house. The smart home automation system shall switch off the lights after detecting no sound or activity for some time.

Self-driving cars are the way to go in future. In future, people don’t have to hire a chauffeur. You may just board on the car and tell it where to go. The car shall use GPS to find the optimal route and take you there, cruising through the traffic. You can get down from the car, and then the car will find an empty parking spot for itself. While you are leaving, call the car from your smartphone to be in the entrance in 5 mins. When you reach out, the car will be waiting for you with the rear door open and playing your favourite music, setting the appropriate temperature for your comfort. Awesome!! Not only that, come next the smart fridges. You can check whether there are beer cans in the fridge by sending a message to the fridge. The fridge shall detect that there are only few beer cans left, and then it can send a message for ordering beer cans from the nearest supermarket. The beers shall be delivered in your home without your involvement. Maybe we can term this as “Self-Replenishing Fridges”. The advertising gimmick for this product would be “The fridge that never becomes empty”.

I may sound like a guy who loath technology and who doesn’t want technology to enhance our lifestyle. However, I believe that I am looking at things with a more conservative viewpoint. In my view, most companies are struck in a red ocean, in which the only focus of companies is to create competitive advantage. The companies’ pursuit of doing things better than the competitors tend to expand the chasm between the technology and customer utility.

I agree that future of technology is all about convergence and integration, i.e, the seamless migration from one environment to another for the end user. In a typical day, people spend majority of their time in three environments: home, office and travel. I believe technology is about integrating all three environments together, so that when a user moves from one environment to another, the devices are aware of the movements (contextual awareness) and take appropriate actions. For e.g, when I move from my home network to my office network, my smartphone shall hide my personal profile and show my official profile. I believe consumer electronics giants such as Apple, Samsung have intentions of dominating in each of these environments. Apple has already penetrated our home with iPod, iPhone and iPad. With the tremendous success in the home segment, Apple extended its offerings to penetrate the other two environments with enterprise (BYOD offerings) and automotive (IoS in car). Once a firm has dominant position in one environment it is easy to offer complementary offerings to penetrate other environments. Similarly, Samsung is also attempting to penetrate the enterprise segment with its Samsung Knox offering. I may have missed to notice any tangible drive by Samsung in the automotive segment. However, the eco-system of Android shall work out in favour of Samsung as most of its smartphones are based on Android. With the growing prominence of automotive apps, Samsung is in a favourable position to also penetrate into vehicle with its android smartphones. I believe rather than embedded intelligence in vehicles, it is more beneficial for both the automotive OEMs and end users to extend the capabilities of the smartphones by using automotive apps. This strategy decouples the mismatch in the product life cycle of automobiles and consumer electronics technology.

It is obvious that tech firms will keep on innovating and create new trends to penetrate in each of these environments. It is also infallible that we consumers will become prey of these tech trends and depend more on these machines instead of our brains. Then I really question whether these smart things are really making us smarter or just offer us an illusion that we are becoming smart.

ARM vs Intel – The way further

In any market, firms compete for the limited market share and revenue. The payoff or returns on investment of the firms are influenced by the exogenous factors such as innovation in related technology, consumer preferences, macroeconomic conditions etc. These factors are beyond the control of firms; however, these have a direct impact on the sales and growth of the firms. Further, as firms compete for the limited market share and revenue, then the future of each firm growth is not only judged by its actions but also its competitors’ actions and reactions. In an idealistic & theoretical setting, each firm would have taken actions to maximize its returns without considering the actions of competitors and the market conditions. Both the market conditions and the competitors’ actions are unpredictable. Each firm takes some actions based on its current understanding of the competitors and market conditions. So, each firm act under bounded rationality as the dynamics of the market conditions and competitors’ actions cannot be predicted properly. Market conditions are driven by macroeconomic factors and consumer sentiments. The firm can only observe the current market state and make some guess about the future trends and then take some action in the current period. In each time period, firms attempts to take the best action, so as to maximize its long term returns on investment. There can also be some effect from the adjacent markets as well. For example, the change in price of tablet may impact the buying trend of ultra-books, in case these are substitutes. At any given point of time, firms act based on the current trends and some future expectations. In the next time period, firms observe the outcome of their actions, competitor’s action in the previous time period and the changes in market conditions, and then take action for the current time period. Thus, firms act, learn and adapt in the market. In the long run, each firm desires to maximize its rewards or profits. However, as the market expansion is limited, the conflicting objective of each firm leads to friction. Many eminent economists and mathematicians have extensively studied such competitive settings using game theory models and infinite horizon discounted cost criterion and then suggested the optimal actions and outcomes of the competing firms. In this post, I shall attempt to explore the competitive strategies adopted by firms in a market using the case study of competition between ARM and Intel in the microprocessor market.

Each firm possess certain set of capabilities that it can execute better than other firms. Sometimes it is also mentioned as core competency. These competencies push a firm away from the competitors and the firm creates a competitive advantage by differentiating its product or services from others. However, in course of time the competitive advantage based on certain set of core competency dilutes as competitors learn about this advantage and imitate it. Then, firms must move on to explore some other form of competitive advantage. Intel is vertically integrated and it has a strong presence in the PC and server segment of microprocessors market. The core competency of Intel is technological advantage as it has a long learning curve of chip design and access to fabrication unit. It can pursue shrinking the process nodes currently from 14 nm to 10 nm and maybe further. It leads the race in setting up the technological trends in the microprocessor market and then others may follow up. ARM with it fabless model of business has core competency in the ecological business model that creates network effects. As ARM supplies only microprocessor IPs, it offers flexibility to the OEMs to customize their chips as per their requirements. Samsung, Apple, Qualcomm have architectural license from ARM and thus they can modify their SoCs as per their needs. In terms of business model innovation, ARM drives the microprocessor market.

However, as mentioned earlier competitive advantage can be replicated by competitors unless those are protected by intellectual property. Intel is entering the smartphone microprocessor market aggressively, in which ARM has dominant share. ARM is also creating an eco-system from last few years to enter the server market, in which Intel has a dominant share. Going further, ARM and its partners intend to catch up with Intel in terms of process node technology going from 22 nm to 14 nm. Intel already has fab ready for 14 nm production; however, ARM partners are currently planning to reduce the process node. In terms of business model, Intel cannot completely follow ARM as the former is vertically integrated. However, Intel can do some changes around its business model to replicate something of the ecological business model. Recently, Intel agreed to manufacture ARM based chips for Altera on 14 nm node maybe to increase its fabs utilization. Further, Intel is also planning to customize its high end processors, so that clients can customize their chips. I believe Intel is slowly realizing the benefits of an ecological business model and it is attempting to pursue some options in this regard without compromising its vertically integrated business model.

Assuming that both the firms exist forever, then what would be outcome of this competition? Can we simulate this competition within the framework of game theory and then derive the optimal strategy and outcome? Last year, while defending my masters in business degree, I pursued this work as my comprehensive project. In the framework of the game, ARM and Intel were the players having some actions based on their core competencies and aspirations. Further, the exogenous impact of the market conditions was taken into factor by various market states: Bull, Bullish, Average, Bearish and Bear. Then, the game was iterated for large number of periods. In each time period, both the firms take some actions based on their current understanding of the other firm’s prior actions, current market state and the expected reward. The firms tend to take actions by which each of them can simultaneously achieve the optimal reward. A further reading on Nash Equilibrium is needed to understand the optimal payoff criterion. In each time period, the firms visit a state-action pair and each of them gets some reward. The next market state is picked up from a transition probability matrix. As the game progresses, the competitors’ knowledge about each other matures and they play rationally with less ambiguity. Finally, with this approach the cumulative payoff for each firm flattens up with no increase in subsequent periods. Thus, the market reaches equilibrium.

However, I am still wondering whether competitive effects can be modelled within the scope of theoretical framework. I believe most theoretical frameworks give good understanding of competition retrospectively. There are many variables that cannot be captured in present period: changes in customer purchasing powers, macroeconomic factors, actions of current competitors, new entrants etc. My work was limited to the quantitative simulation of the competition in the microprocessor market.

The question still remains: what is the way further in the competition between ARM and Intel considering each firm last forever? Is market equilibrium possible in this scenario and what would be the equilibrium strategies for both the firms? Is it possible to make out which firm shall win in the long term? I am not an expert in competitive market strategy; however, with my limited understanding I can just mention that each firm shall leverage its core competencies to expand their business horizon in new domains. Each firm shall jump from one emerging trend to other emerging trends to keep up pace with the market and competitor. It is impossible to derive an optimal strategy for each firm as there can be no fixed recipe for succeeding under uncertain market conditions and competitive effects. So, the strategy that each firm pursue shall be mostly emergent rather than formal in order to factor in the unpredictable market scenarios and competitive actions.

Currently, smartphones market is exploding. Many new OEMs and SoC designers are entering the market to cash on the demand. As any market becomes attractive, competition increases and thereby deteriorating the profits of competing firms. Smartphones market is already crowded with multiple offerings and OEMs struggle to differentiate their offerings from other similar products in the market. Further, there is also impact from the adjacent markets. Wearable devices are coming up and in future some of them may replace smartphones. Each firm in the smartphone market intends to differentiate their offerings. The differentiators come in areas – hardware, software, supply chain, post-sale services, disposal etc. I shall limit the scope to only hardware and software. I believe it is easier to replicate software; however, differentiation in terms of hardware will need major investment in capital and knowledge. Fabrications units such as TSMC will have to invest heavily migrating to 14 nm or further down in process technology.  ARM, Intel and few other firms such as MIPS have aspirations of dominating the smartphone microprocessor market. ARM has to struggle to maintain its dominant share with its business model innovation. Intel strives to crack ARM’s domination with technology revolution. In terms of exchange of best practices, ARM and its partners can catch up with Intel’s technology with some lead time and capital investments; however, Intel may struggle to follow up the business model innovations of the former. Currently, differentiation in smartphone segment is more limited to technology push such as true octa-core, BIG:LITTLE, symmetric processing etc; however, with adoption of smartphone saturating in the developed nations, the next big market will be developing nations such as China, India. I can assert that Indian customers seek value for money. Products with most utilitarian offering and least price shall usually crack the market. This is driving many Indian OEMs market share such as Micromax, Lava, Karbonn etc. However, price war is not an optimal answer to competition. The challenge is to pursue price reduction and customer value creation simultaneously. The current factors of competition revolve around price, battery life, performance, apps. So, it is difficult for OEMs to differentiate their offerings in the smartphone market. In such scenario, what Intel and ARM would do, so that both the price comes down and end-users value creation occurs? I believe that technology push cannot penetrate much in Indian market; a new factor of competition has to evolve. It may happen in future, we may not need high performance smartphones at all. The phone will be a stripped down version of current smartphone with excellent connectivity network. All the processing will be done in a remote cloud server. The phone shall just act as a virtual desktop providing remote connectivity to the cloud server. The memory requirement in smartphone will reduce drastically as most of the stuff shall be offloaded to the private cloud. I believe that technology enhancement may not be the only factor of competition in smartphone market. Firms should explore using technology to address customer experience in an economical way. Else, feature phones may still continue to rule the mass market in developing nations.

Both ARM and Intel are well positioned in the microprocessor market in terms of their capabilities. I believe the way further in this competition should be inclined towards understanding new emerging market trends and addressing the user needs in an economical manner. Technology push may not decide the dominant firm in long term.

My post is an attempt to understand the complex reactions and actions that firms take in a market space to gain revenue and market share. I tried to explore this using the case study of microprocessor market competition. I am not an expert either in microprocessor domain or in competitive market strategy. I just write my views with my limited understanding. I would really love to hear comments on my post.