Disruption of multi-decade-old global value chain & the key trends that we are excited about

Authored By: Nesara S R & Abhishek Singh

This is the second part of the article on the state of affairs of semiconductors, the previous part could be accessed here.

Disruption of multi-decade old global value-chain: Key tailwinds driving growth in the semiconductor industry

2.1 Technological advancements

AI, Digitization, Leading-edge demand and more

The rise of AI and its applications, particularly in generative AI, has created a surge in computing requirements. This has led to the development of specialised chips designed to enhance neural network performance and accelerate AI workloads. The growing digitization trend, amplified by the COVID-19 pandemic, has further propelled the chip industry. As the world shifted online, there was a substantial demand for consumer electronics such as laptops, smartphones, and other connected devices. Additionally, the need for robust network infrastructure, including Wi-Fi routers and network devices, skyrocketed as more people relied on digital connectivity. The combined effect of AI advancements and digitization has created a lucrative ground for innovation and growth in the chip industry, driving the development of cutting-edge technologies to meet the evolving demands of a digitally connected world. The pursuit of higher performance and energy efficiency in various applications, including AI, data centres, and high-performance computing, has also intensified the need for chips tailored for desired specifications. Take Apple, for instance. In part 1, we classified them as an OEM. However since 2020, Apple has also ventured into becoming a fabless design company! Realising that Intel couldn’t keep up with their pace of innovation, Apple started designing their own chips. You’ve probably seen the impressive M1/M2 SoCs in the recent Apple MacBooks and PCs. Similarly, in 2021, Google made headlines with the introduction of custom server chips specifically tailored for YouTube video processing. These chips delivered an impressive 20x boost in efficiency, enabling faster and more seamless video streaming experiences for users. Not to be outdone, AWS in 2022 unveiled their own custom-designed Graviton chips. These chips were specifically developed to optimise performance in AWS data centres and offered a remarkable 50% better price-performance compared to the prevailing state-of-the-art solutions.

The pursuit of higher performance and energy efficiency in various applications, including AI, data centres, high-performance computing, IoT etc has intensified the need for chips manufactured using these leading-edge nodes and has prompted chip manufacturers and design centres to invest heavily in R&D.

Open-source disruptions

In the world of chip architecture, there are different styles or “rules” that dominate the industry. The most well-known and widely used architectures are x86 (by Intel), ARM (by ARM Ltd), and the relatively new RISC-V.

Historically, the PC market has been dominated by the x86 architecture. Apple’s decision to transition to their own ARM-based processors for their Macs has brought some change, although Windows-based PCs still primarily use x86 processors. When it comes to data centres, x86 architecture remains the dominant choice. x86 was specifically designed with cloud computing in mind, offering the performance and scalability required for these environments.

On the other hand, ARM architecture has been a standard in the industry, but it is a closed and IP-protected architecture owned by ARM Technologies, a Softbank-owned company. ARM charges royalties to its partners who incorporate ARM’s intellectual property into their chips. The licensing fees can range from an estimated $1M to $10M, and the royalty is typically around 1–2% of the chip’s selling price. This dependence on a single tech company, which by the way has a volatile history, with the industry often labelling it as an “unstable” partner and the associated costs that come with licensing have come to be highlighted as major limitations for ARM.

In contrast, RISC-V presents an interesting opportunity. It is an open-standard architecture, allowing anyone to build products based on it. Earlier this year in Jan, Google announced RISC-V architecture for its Titan M2 security chip in the Google Pixel series. Apple also announced that the Mac platform would migrate away from the ARM-based M1 & M2 processor line, in favour of the RISC-V architecture. Tech giants have already started investing in RISC-V opportunities. Amazon also seems to be getting in line to tap into the RISC-V potential although the exact end use case where it will be deployed is yet to be known.

With companies increasingly becoming eager to own their tech, there is no argument against the lower development costs RISC-V can result in due to the absence of licensing fees and its open-source nature. This eliminates a significant barrier in chip development.

Although RISC-V is still evolving and not yet in its optimal form, many experts believe it could be the next big thing in computing for the next decade. Its flexibility, customization options, and potential for innovation make it an attractive choice for various applications and industries.

There have been similar promotions of open source and standardisation in the automotive sector as well. AUTOSAR (Automotive Open System Architecture) is a standardised automotive software architecture that aims to enhance interoperability, scalability, and reusability in the development of embedded systems for vehicles. It provides a common platform for automotive manufacturers, suppliers, and developers to collaborate and streamline the design, integration, and maintenance of software components. This contributes significantly to the emerging areas of the automotive industry such as connected autonomous vehicles.

The key takeaway here: With the advent of open source and reduced costs in the semiconductor industry, there is greater scope for innovation. Companies can focus on developing proprietary intellectual property (IP) in specific areas while leveraging standardised and open-source components for other parts of their designs and customising it for their specific use cases, enabling faster development cycles, improved interoperability, and cost efficiencies in the creation of chip solutions.

2.2 Geopolitical developments

Growing volatility with China

Taiwan stands as the largest supplier of chips worldwide, playing a pivotal role in the semiconductor industry. While chip design, assembly, and testing can be conducted elsewhere, Taiwan serves as the epicentre for major foundry activities. Notably, Taiwan’s semiconductor production accounts for over 60% of the global supply, including a majority share of the most advanced chips as we saw in Part 1 as well. However, the delicate relationship between China and Taiwan, marked by cyber-attacks, constant tension and threats of invasion poses a significant risk. Any disruptive actions by China could bring the world’s semiconductor market to a standstill, which would have far-reaching implications. To put things in perspective, TSMC’s factories faced temporary shutdowns during the COVID-19 pandemic, contributing to a chain reaction of supply deficit leading to the global chip crisis in 2021. In the event of China’s invasion of Taiwan, there is also a risk that it could gain access to leading-edge node-size foundries, thereby bolstering its military and defence capabilities.

This situation underscores the critical importance of reducing dependence on a single company or country for chip manufacturing. Efforts have been made, such as the US persuading TSMC to pledge a $12B foundry in Arizona, but further diversification is crucial. Even beyond Taiwan, the concentration of chip supply around the South China sea makes the rest of the world uneasy. Building a resilient and distributed global chip manufacturing ecosystem is essential to mitigate risks and ensure the industry’s long-term stability!

Nations striving for “chip independence”

Countries around the world are recognizing the critical importance of diversifying their chip manufacturing capabilities. To achieve this, both independently and collaboratively, they have embarked on initiatives to enhance their core competencies in the industry. One notable example is the Quad Summit, a gathering of India, the United States, Japan, and Australia that has been taking place annually since 2021. One of the key objectives of this summit is to collectively address all aspects of the chip value chain within these four countries, thereby reshaping the global supply chain and reducing dependence on China’s actions. The US signed a “Chips Act” in 2022, providing roughly $280B in new funding to boost domestic research and manufacturing of semiconductors in the United States. The EU signed its own Chips Act to secure its position by aiming to attract $43B in public and private investments. To put this into perspective, it is noteworthy to mention that China has already invested a significant sum of $100B to date, and they have additionally pledged a staggering $220B for future investments.

Furthermore, India and other South Asian countries like Thailand, Vietnam, and Cambodia have emerged as attractive alternatives in the “China plus one” strategy adopted by other nations. Foxconn, a leading contract manufacturer, is transforming India into its secondary manufacturing base by establishing production capacity for EV parts and Apple iPhones. Additionally, Micron has recently committed to setting up an Outsourced Semiconductor Assembly and Testing site in Gujarat, India. India itself has ambitious plans to bolster its in-house foundry ecosystem. In December 2021, the Semicon India Programme was approved by the cabinet, allocating Rs 76,000 crore for the development of the semiconductor and display manufacturing ecosystem in the country. As part of recent modifications to the program, the government has increased the financial incentives for companies, consortia, and joint ventures by providing 50% of the project cost for setting up semiconductor fabs of any node size. The government is actively pursuing its endeavours to modernise and commercialise India’s sole semiconductor chip fab, the ISRO-owned Semiconductor Lab (SCL) in Mohali, through a joint venture. Vedanta is already in fact exploring the possibility of availing this incentive scheme through a joint venture with Foxconn. The facility boasts full production lines of industry-grade equipment, specifically catering to the 28 & 40 nm node sizes. Similarly, as per reports, a UK-based company is planning to set up memory-chips manufacturing facilities in Odisha. Given India’s anticipated fabrication capabilities, the country is well-positioned to focus on lagging node sizes. Furthermore, the SCL foundry is expected to soon enable the production of compound semiconductors (GaAs/GaN) as well.

The key takeaway here: Now would be a good time to gear up to (1) support this growing ecosystem in India as well as (2) connect India with various other emerging hubs across the globe.

Emerging themes: The business models that we are most excited about

In our view, entering the semiconductor industry with a capital-intensive model such as outsourced manufacturing or assembly solutions right from the early stages may not be viable, especially from a VC fund’s investing life cycle perspective.

Therefore, we believe it is crucial to approach the semiconductor industry with a strategic and differentiated perspective, focusing on areas where innovation, niche solutions, and value-added services can make a meaningful impact and drive long-term success.

Aligned with this line of thought, we have identified a few major areas where the semiconductor space is gaining significant momentum. The time is right for founders in India to venture into these areas, considering the recent developments and initiatives taking place. Government is actively supporting the semiconductor industry through various incentive schemes and policies, creating a conducive environment for starting up. Moreover, technological boundaries are constantly being pushed, creating new opportunities for innovation and disruption. Lastly, these areas offer relatively lower barriers to entry, allowing startups to quickly enter the market, build their solutions, and scale their businesses.

1. Increasing adoption of SoCs as off-the-shelf chips starts to take the backseat

Custom chips and SoCs are gaining more popularity over off-the-shelf ones, and you may wonder why. Well, the demand for these specialised semiconductor solutions is fueled by various factors, including the need for specialised performance, product differentiation, technological advancements, and cost optimization. As industries continue to innovate and push boundaries, the requirement for such specialised use cases is growing too. As we saw in the previous section, giants like Apple, Amazon, Google have all made headlines for embracing the need for custom SoCs.

Now, imagine the scale of possibilities across high yield industries such as automotive, defence, aviation, telecom, and consumer electronics (the list goes on). There’s a whole lot of them that aspire to install chips tailored to their specific requirements. However, it’s not as simple as Apple or Google made it look. Developing custom chips demands extensive R&D efforts, along with a pool of talented individuals with the right expertise.

This creates a lucrative opportunity for founders who can develop solutions that offer accessible design capabilities to OEMs of all scales but particularly SMBs and Enterprises that often have limited influence on large design houses and fabrication facilities with extensive R&D resources. As evidence of this growing potential, India has witnessed a remarkable surge in fabless design startups, with the number expected to reach 100 by 2024, a significant leap from near-zero startups just a year and a half ago. The semiconductor industry’s total value addition, according to this report by BCG, amounts to $290B. Of this, a substantial portion of $150B (~50%) is being contributed by design, core IPs, and EDA (electronic design automation), despite commanding a mere 13% share in CAPEX across the entire value chain. This ratio of value addition to capital expenditure also makes this sector highly attractive. In such a case the possibilities are immense :

1. Design infrastructure & software solutions that simplify the chip design process and make it more accessible to OEMs. These tools can include user-friendly interfaces, automation features, and simulation capabilities, enabling faster and more efficient chip development. In terms of positioning, building an infrastructure on top of RISC-V for enterprises to be able to build customised SoCs for their own requirements.
2. Create and licence IP related to chip design. These can be innovative designs, architectures, or algorithms and offer them to OEMs as licensable IP, allowing OEMs to incorporate these cutting-edge technologies into their own custom chips.
3. Niche expertise in specific areas of chip design such as power optimization, memory design, security/cryptography, AI-accelerated chip design etc
4. Design services and consulting to offer expertise and support throughout the chip design process. This can include assisting with architecture design, layout optimization, testing, and validation.

By addressing one or more of the aforementioned use cases, significant value can be created by assisting OEMs in simplifying the chip design process.

2. Digitising Electronic Manufacturing Services

Cloud EMS (Electronics Manufacturing Services) providers operate without the need to own or operate their own manufacturing facilities. Instead, they leverage a network of vendors to offer electronic manufacturing services. In this model, Cloud EMS providers would act as intermediaries between OEMs and vendors. They facilitate the entire electronic manufacturing process, from offering consulting support to the R & D team in design and prototyping, BoM (Bills of material) procurement to PCB fabrication and assembly, by coordinating with their network of vendors.

The cloud EMS model brings benefits to all the players in the ecosystem. Traditionally, the industry has been highly fragmented:

1. Many PCB assembly vendors operate at around 40–60% capacity utilisation limiting their ROCE.
2. Similarly, for each electronics product that on average needs 500–100 components, the OEM/EMS are expected to coordinate with 60–80 vendors distributed across the geographies, the experience gets further inefficient given the lack of active components vendors domestically, and
3. Lastly, with the lack of PCB fabrication capacity in India, the discovery of vendors by OEM for their requirements is yet another challenge given the matching that needs to be done based on the capability of the vendor and the number of fabrication layers required by the OEM for their product.

However, the introduction of cloud EMS intermediaries addresses this issue by acting as a bridge between supply and demand. These intermediaries have access to a wide network of manufacturing partners with diverse capabilities and expertise. By leveraging their network, they can effectively match the specific requirements of customers with the most suitable manufacturing partner.

On the demand side, especially for mid-sized and small enterprises, cloud EMS offers significant benefits. These enterprises may not have the resources or expertise to navigate the complex landscape of the traditional EMS industry. Cloud EMS allows them to tap into high-quality manufacturing services without the need for large-scale investments or extensive supplier research.

3. Bridging the Gap between OEMs/Fabless and Emerging foundries

Large-scale vendor networks strive to keep OEMs/Fabless well-informed about the latest updates on pricing, quality, innovation and other trends. However, there remains a blind spot when it comes to the innovation happening on the active components with smaller emerging fabless/foundries. With hundreds of players across the globe, it becomes impractical for OEMs/Fabless to scan and discover every potential opportunity. This is an area where value can be added by building an asset-light model that specialises in this discovery and matchmaking process. The current landscape presents a unique opportunity for founders, driven by the tailwinds mentioned in the previous section of this article:

1. Heavy semiconductor manufacturing capacity expansion was triggered in the US and Europe in response to the global chip crisis that occurred post 2020. Added to this, as highlighted in the previous segment, we can expect to see more semiconductor capabilities shaping up in newer regions as well. This expansion creates alternatives in the supply chain, especially for MSMEs heavily reliant on Southeast Asian distributors and stockists, predominantly Chinese. The “China plus one” strategy to diversify the supply base necessitates solving the discovery challenge for the new capacity that has emerged.
2. The existing Chinese distributors and stockists form a long chain with limited transparency in pricing and quality. By replacing a few steps in this chain and providing customers with more options, you can deliver significant value.

We are therefore excited about marketplace models that can emerge with solutions to consolidate the supply-side, their technologies, and products, making it easier for OEMs/Fabless to discover and evaluate potential partners. Even better if these marketplaces are supported by data-backed frameworks to connect OEMs/Fabless with semiconductor companies based on their specific needs and requirements. Moreover, in the long run, this gives an opportunity to build long-term relationships with customers and eventually integrate vertically to provide a large suite of services.

4. Supporting energy efficiency aspirations

Semiconductors play a crucial role in enhancing energy efficiency across various industries. The Complementary Metal Oxide Semiconductors (CMOS) technology has proven to be well-suited for low-power devices, however, for high-power applications, the utilisation of compound semiconductors such as GaAs and GaN holds immense potential due to its intrinsic material properties. These compound solutions are capable of cutting power consumption by upto 80% as compared to silicon-only alternatives. Sectors like electric vehicles (EVs), EV charging systems, and telecom stand to benefit greatly from the adoption of these advanced materials. As highlighted in the previous section, the SCL in India is actively enhancing its foundry capabilities to facilitate the production of GaAs and GaN. Furthermore, the advantage lies in the fact that high-power applications can effectively leverage lagging node sizes ( ≥ 65 nm), which are more accessible and readily available compared to leading-edge technologies. Thus, securing manufacturing resources also becomes relatively easier due to the relatively more widespread availability of lagging node-size foundries.

A model example in this space is the US-based vertically-integrated company Transphorm, which successfully achieved market readiness in 2019 with mass production capabilities and even went public with an IPO. Within 3 years of operations, the company closed FY23 at ~$16.5M in revenue and is valued at ~$200M. It also has a promising revenue pipeline of $440 million projected over the next five years.

India, recognizing the significance of this technology, has a tremendous opportunity for Fabless companies in this area.

5. Human resource development for the semiconductor industry in India

While the establishment of foundries in India is on the horizon, one key challenge lies in the lack of experience among the country’s young and talented individuals in working within foundries. Bridging this gap by developing the necessary skill set through comprehensive training and skill development in collaboration with companies that possess existing foundries, & eventually enabling job linkages/matching becomes imperative.

I think that’s it for now folks! I hope that this post was helpful to the readers. I would love to hear any comments and feedback on the above. And, if you’re building in semi-conductor space, or are a VC actively following this space, please don’t hesitate to get in touch. My email is abhisheksingh@riverwalkholdings.com. I would love to connect with you to brainstorm and identify the overlaps.

Additionally, we are maintaining a repository of all the semiconductor start-ups in India here. Please feel free to reach out if you would like to suggest a startup for inclusion in the list or propose modifications to any entry.