When Semiconductor Manufacturing Went East and Chip Design Followed

The same cost and talent forces that once shifted chip manufacturing to Asia are now moving semiconductor design out of America, making India a powerhouse of design talent.

Today’s post is fully free and discusses why India is emerging as the chip design capital of the world. If you’re new, start here! On Sundays, I write deep-dive posts on critical semiconductor technology for the AI-age in an accessible manner for paid subscribers.

I also post video content on YouTube based on articles from the newsletter, and ideas that fit a verbal format better.

---

Everything in its right place

It is trite to say that semiconductors is a global business. Using European wafer fab equipment and chemicals from Japan, chips are manufactured, packaged and tested in Southeast Asia. Supply chains are global as we found out the hard way during the Covid era; this has recently led many countries to try and develop chips indigenously.

We are here today because various countries had unique “strengths” which allowed them to define their space in the semiconductor industry. Outsourced Semiconductor Test and Assembly (OSAT) has been dominated by Southeast Asia since the 1960s primarily because they provided services at an order of magnitude lower than American chip makers could possibly manage in the US. As Chris Miller writes in Chip War,

In 1963, its first year of operation, the Hong Kong facility assembled 120 million devices. Production quality was excellent, because low labor costs meant Fairchild could hire trained engineers to run assembly lines, which would’ve been prohibitively expensive in California.

Silicon Valley’s famous taskmaster, the late Charlie Sporck, soon expanded Fairchild’s production to Taiwan, Malaysia, Singapore and South Korea because the cost of labor was even lower than Hong Kong, and most importantly, because there were no labor unions in Asia burning effigies of him for overworking them.

Morris Chang founded TSMC with the help of the Taiwanese government after getting shot down by Texas Instruments for proposing a manufacturing-only business unit without competing on the designs being made (and getting passed over for the CEO role). When Carver Mead and Lynn Conway described a set of “design rules” which, if followed, meant that anyone could design chips, it was a “Gutenberg moment” for semiconductors, as Miller puts it in his book. The fabless company model was born.

Who actually designs America’s chips?

The design of chips themselves, though, has historically been an American specialty. The largest chip companies in the world - AMD, Nvidia, Apple, Texas Instruments, Qualcomm - to name a few, have operated successfully on the fabless model. But, a closer look at the workforce in these American fabless companies reveals a different story.

Apple’s design is proudly Californian.

The chart below shows the results from a Bureau of Labor Statistics (BLS) study from 2017. The study looked at the ratio of native-born to foreign-born labor force in metropolitan IT companies from 1980 to 2014. With every passing decade, foreign born engineers have evolved to occupy a larger fraction of the workforce. Right about twelve years ago, foreign labor took a majority in the IT industry. This includes both software and hardware, but the ratios should broadly apply even if just looking at people working on chips. Extrapolating to present day, foreign labor dominates the technology industry. Lots of handwaving of course, but its the trend that is the message here.

US civilian labor force, creative computer occupations in IT industries in the Metropolitan areas with innovative IT clusters, by nativity, 1980, 1990, and 2000-14. Source: ViksNewsletter; Data: Bureau of Labor Statistics (link).

The only way America has been able to sustain the level of foreign-born talent is through its H1B visa program, which has of late been under attack causing people to abandon air travel. The suddenly imposed $100,000 for a new H1B visa increases the cost of hiring fresh graduates by 2x with the added uncertainty of even getting a spot in the limited quota of H1B visas available. This leaves US tech companies with a few choices:

1. Pay the increased fee and keep the talent pool alive while taking a hit on revenue.

2. Oppose the administration’s policies on immigration, risk getting hit by tariffs on products and access to US power infrastructure that is getting increasingly competitive with datacenter buildouts.

3. Accept the policy, limit the number of H1B hires, and instead off-shore jobs to other countries with a large pool of available talent at a lower cost per employee.

It does not take a fortune 500 CEO to make the reasonable choice here. And, I’ll steer clear of the politics involved in the current situation.

India as the Fabless Design Capital of the World

This storm of events has opened the doors to India becoming equivalent to what Southeast Asia was in the 1960s: a place to outsource chip design at low cost - relinquishing America’s last stronghold in chips: design.

Today, India is well poised to take its spot as the fabless design capital of the world.

A large fraction of the tech workforce of a company is already from India whether they are physically working in the US or India. As a country, India is predominantly English-speaking, produces about a million engineering graduates each year, and provides a talent pool that performs the same engineering roles at a fraction of the US cost. Additionally, India’s politically neutral stance with the US makes it a better choice compared to China, which also produces a copious amount of engineering talent each year.

Several major American chip companies already have a significant footprint in India - in the major metropolitan cities of Bengaluru, Chennai, Hyderabad and Noida. For example, Intel India has 13,000 employees accounting for about 15-20% of the total workforce and is the largest design center outside the United States. Nvidia has anywhere between 4,000-10,000 employees (10-30%) of its workforce in India. A snapshot from Cadence’s 2024 annual corporate impact report shown below says that India has an even higher percentage of employees than in America. Another company I cannot talk about (legal pitchforks 🔱) has a significant presence too. The off-shore presence of these companies is only getting larger.

Try this: for any company of interest, look up the number of job listings in the US versus other countries and let me know what you find in the comments below.

Fabless Expertise vs Silicon Sovereignty

While US-based companies do have a significant footprint in India, the globally prevalent theme of natively developing silicon technologies to reduce dependence on outside nations is definitely present in India too. In the recent Semicon India conference held in New Delhi, the Vikram Sarabhai Space Center announced the Vikram-32 microprocessor manufactured by the Semiconductor Laboratory (SCL) for satellite applications. It uses a custom instruction set architecture and utilizes a 180 nm process node. This chip was hailed as a fully “make-in-India” silicon effort to demonstrate India’s indigenous design capability.

There are two major problems with establishing silicon sovereignty this way.

• Making chips is a global affair - like it or not. Just developing chips does not reduce dependence on other countries. Chips need to be packaged, assembled, tested, and sold globally.

• A 180 nm node is 25 years behind state-of-the-art. Getting to leading technology nodes requires enormous capital investment and unique expertise - as we are seeing with Japan’s attempt at a 2 nm node with Rapidus. More likely, India will at best target a 28 nm node for analog/RF/mixed-signal applications.

A better approach for India is to double down on fabless chip design expertise instead of trying to own the soup-to-nuts of the chip design flow. Fundamentally, India faces three fundamental problems: capital expenditure, experience, and competitiveness.

I watched an interesting fireside chat at Semicon India 2025 in which Dr. Naveen Yanduru, CEO of Axiro Semiconductor (a spinoff of Renesas’ RF BU that has recently headquartered in Bengaluru, India) made some interesting suggestions to solve these fundamental problems, which I paraphrase below.

1. Expenditure: Chip design is an expensive process requiring EDA tools, IP licenses, mask costs and testing equipment. At present, this is the buy-in to play at the global semiconductor table. However, the rise of AI-enabled EDA might offer enhancements that both lowers costs, and lowers the high knowledge barrier to chip design with better tooling.

2. Experience: Working for companies that already have a global footprint is the best way to develop chip design expertise and a deep understanding of the challenges involved in making and selling chips at scale. American companies and Indian companies with a global presence provide plenty of opportunity.

3. Competitiveness: There is no shortcut to this. Developing the best competitive products takes time and effort. With sufficient financial investment and mentorship, new startups can be encouraged to find innovative problems to solve instead of trying to beat existing semiconductor giants at their own game.

While the fireside chat concluded with defining a metric for India’s semiconductor success as “building a fully-native chip, with Indian made silicon, packaging and test,” I’d argue that the success metric be defined differently.

Build a fabless Indian semiconductor unicorn - a company with a valuation of $1 billion - with products that are competitive on the global semiconductor stage before 2035.

You can watch the whole fireside session from Semicon India 2025 below. Specifically, the 45 minute segment starting from 3 hour, 14 minute mark.

---