Jensen Huang, chief executive of Nvidia Corp., on the earnings call for the chipmaker’s financial second quarter ended 28 July.
A glimpse of this revolution was already visible in Nvidia’s full-year earnings for 2023-24. Its data center revenue more than tripled to $47.5 billion from $15 billion in FY23, accounting for 78% of its overall revenue of $60.92 billion. Net income surged almost sixfold to $29.76 billion.
It wasn’t just the fundamentals, the technical charts also shot up.
Nvidia’s stock price is trading at 98% spread on its 200-day-moving average, surpassing Cisco’s record of 80% spread on its 200-DMA in March 2000. That’s the wildest spread a US firm has reported since 1990, noted Jonathan Krinsky, chief market technician at research and broking firm BTIG.
View Full Image
What made Nvidia a global household name is the ChatGPT-led generative AI revolution. Its stock price has jumped 700% since the generative AI frenzy began in November 2022, propelling a company that had a market capitalisation of $350 billion in November 2022 to a $3.3 trillion valuation in June 2024. The Nasdaq Composite Index only grew 65% in this period.
Nvidia beat records by selling shovels in the gold rush, and oil in the industrial revolution.
Indian conglomerates enter the semiconductor space
While the world changed for Nvidia post-pandemic, it also changed for India’s semiconductor landscape. After multiple failed attempts at making semiconductors in-house, India is now taking lessons from history and firing all cylinders to build a semiconductor ecosystem.
The SEMICON India conference running from 11-13 September is a testament to how serious the government is about building the world’s most complex piece of technology.
Government financial support, favourable policy, and technology collaborations with the US, the UK, Japan, and Singapore have attracted deep-pocketed conglomerates including the Tata group, the Adani group, and Larsen & Toubro Ltd to venture into the semiconductor space. These conglomerates do not have the technical know-how to build a chip. Yet they are making huge bets on chips in partnership with other foreign chipmakers who have the know-how.
This brings us to this question: Can India’s conglomerates give Nvidia a run for its money?
For more such in-depth analyses, read Profit Pulse.
Friends before enemies
We first need to consider what Nvidia has that Indian companies don’t—the intellectual property to make the most advanced graphics processing units (GPUs).
Nvidia invented GPUs. Intel Corp., Adanced Micro Devices Inc. (AMD) and many other companies tried to make GPUs but couldn’t beat Nvidia in its game. This success didn’t come to Nvidia by chance. It was a tireless effort of 30 years, product failures, and a vision for accelerated computing and AI. Nvidia envisioned AI back in the mid-2000s when AI was just a thought.
Nvidia built a whole AI infrastructure around its GPUs, with software support, and use-case applications, making it easier for companies to adopt its technology. Today, Nvidia’s GPUs power every AI-enabled data centre, which Huang calls “AI factories” that produce “a new commodity: artificial intelligence”.
View Full Image
In fact, Tata group and Reliance Industries Ltd have partnered with Nvidia to build India’s AI infrastructure. When Huang visited India in September 2023, he had a dialogue with Prime Minister Narendra Modi on developing AI skills in the front office (or engineering, marketing, sales, finance, and business operations) where the technology can create value for companies.
During that visit, Nvidia partnered with Reliance to co-create AI solutions. The outcome of this partnership was visible in Reliance’s latest annual general meeting, when chairman Mukesh Ambani announced a gigawatt-scale AI-ready data center in Jamnagar and Jio Brain (a comprehensive suite of AI tools and platforms to streamline AI adoption).
Nvidia also partnered with Tata Motors Ltd, Tata Consultancy Services Ltd, and Tata Communications Ltd to build AI infrastructure, accelerate the build-out of AI solutions, and reskill 600,000 IT professionals of TCS on developing front-office AI applications.
Instead of competing with Nvidia, India is befriending it to catch up in the AI race.
Also read | Central banks are betting big on gold. Here’s how to profit from the trend
The other side of the chip
India’s business houses are also injecting crores or rupees into building chips in India.
The global semiconductor manufacturing landscape has a very tight supply chain, where a handful of companies make up for most of the chip-making. There are chip designers like Nvidia and AMD, material providers like Applied Materials Inc., manufacturing equipment suppliers like Lam Research Corp., foundries like Taiwan Semiconductor Manufacturing Co. Ltd (TSMC) that fabricate chips, and assembly, testing, marking, and packaging (ATMP) facilities, where China leads.
Nvidia doesn’t manufacture its GPUs. It only designs and sells the product, outsourcing the chip manufacturing to TSMC, which has the most advanced fabrication facilities, or fabs.
Nvidia, Apple Inc., Qualcomm Inc. and many other companies have adopted the fabless model to build their niche in chip designing and selling. Believe it or not, India has 20% of the world’s chip designing talent, most of which are small startups.
Tech-to-construction conglomerate L&T entered chip design in 2023, beginning with designing system-on-chip (SoCs) across 40nm (nanometer) and 7nm nodes. It is investing another $300 million to design 15 products, including power chips, radio-frequency semiconductors, and mixed-signal integrated circuits. It has even partnered with International Business Machines Corp. (IBM) to develop advanced processors.
However, GPUs are not a part of L&T’s roadmap as of now.
Tata and Adani’s semiconductor bet
While L&T is starting its semiconductor journey with chip designing, Tata Electronics Ltd is investing ₹91,000 crore to build India’s first fab in Dholera, Gujarat in partnership with Taiwanese foundry Powerchip Semiconductor Manufacturing Corp. (PSMC). This fab will build different types of chips on matured nodes of 40nm and above.
Taiwan’s TSMC and South Korea’s Samsung Corp.—the world’s largest and most advanced foundries—are currently working on building 2nm fabs given the strong demand from Apple and Nvidia. The two foundries are already mass-producing chips on 4nm node. TSMC even produced the world’s first 3nm chip—A17 Pro inside Apple’s iPhone 15 Pro—last year and is building A18 chips for the upcoming iPhone 16 series on 3nm node.
So why are Indian companies building nodes of 40nm—a 2007 technology?
View Full Image
In the semiconductor world, the most complex are processors like central processing units (CPUs) and GPUs, which are the brains that direct other chips. Smaller nodes add value to these chips. However, analog, sensors, radio frequency, and power chips are used in almost all electrical equipment and can be made on mature nodes like 45nm and 65nm without compromising performance.
A mature node is less expensive, generates a high production yield, and can be better capitalised compared to expensive smaller nodes. Tata is looking to capitalise on mature nodes and build different types of chips used in automotive, computing and data storage, wireless communication, and AI.
The ports-to-power conglomerate Adani group is investing ₹83,947 crore to build another fab in Taloja on the outskirts of Mumbai in partnership with Israel-based analog chipmaker Tower Semiconductor Ltd. It will make chips for drones, cars, smartphones and other mobility solutions.
Can Tata and Adani give Nvidia a run for its money?
The two conglomerates have just started their semiconductor journey. Having no technical know-how, they are taking a structured approach to first learn the ABCs of chip-making and make money in the process. This will give the companies some room to experiment and absorb losses from initial failures under the conglomerate umbrella. Foreign partnerships will give them access to technical know-how and upskilling of human capital.
Their initial aim is to make chips in India and reduce reliance on imports. Once they master the mature nodes and develop a market presence, they could consider upgrading to advanced nodes.
Maybe 10-30 years from now, they could also become innovators. At that juncture, these companies would have to decide whether they want to develop a whole new chip like Nvidia (GPUs) and Intel (x86 architecture), or an alternative to existing ones.
Apple has developed M-series processors for its laptops as an alternative to Intel’s CPUs, and Samsung has developed Exynos processors as an alternative to Qualcomm’s Snapdragon.
Can a conglomerate beat a tech company?
Tech companies are out there to change the world. But conglomerates are out there to commercialise the technology. The biggest example of a conglomerate is Samsung.
Founded in 1938 as a trading company, Samsung Group ventured into a plethora of businesses including retail, construction, textiles, and food processing. At one point it also manufactured cars and planes.
Samsung entered the electronics industry in the 1960s and ventured into making electronic devices as well as semiconductors. By 1992, it was the world’s largest producer of memory chips. In 2021, it overtook Intel to become the world’s largest chipmaker by revenue.
View Full Image
Today, Samsung has also become an innovator with its fold technology (Galaxy Z Fold), which took three generations of innovation and a decade of progress to make. It has committed to build nine fabs with a total investment of $373 billion, and help South Korea build the world’s biggest semiconductor cluster by 2047.
The bottom line
India’s semiconductor industry is at a nascent stage, its $10 billion investment just a drop in the ocean. But the momentum that began with the government’s productivity-linked incentive scheme for the semiconductor industry should increase if Indian companies are to compete with the likes of Nvidia.
Samsung’s journey gives investors hope that the Tata and Adani groups could overtake Nvidia in about 30 years. For now, semiconductor investments should not be put through the litmus test of return on investment (ROI) and price-to-earnings (PE) ratios. Instead, these should be looked at as long-term opportunities waiting for an inflection point, just as Nvidia did with generative AI.
Also read | This seasonal trade could work wonders. All it needs is an X factor.
Note: The purpose of this article is only to share interesting charts, data points and thought-provoking opinions. It is NOT a recommendation. If you wish to consider an investment, you are strongly advised to consult your advisor. This article is strictly for educative purposes only.
Puja Tayal is a seasoned financial writer with over 17 years of experience in the field of fundamental research. In her articles, she brings a good blend of comprehensive, well-researched insights into a company’s workings.
Disclosure: The writer or his dependents may or may not hold the stocks/commodities/cryptos/any other asset discussed in this article.
