Jan 24, 2025 · 1h 6m · neon-show
Nvidia’s Success, Chip Race, India’s Semiconductor Mission, & Hardware Vs Software | Raja Manickam
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The Neon Show, host Siddharth Ahluwalia interviews semiconductor pioneer Raja Manickam to explore the structural resurgence of hardware, NVIDIA's architectural dominance, and India's strategic push to build domestic semiconductor sovereignty and fabless innovation.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
Siddhartha as informed peer 3.6 Guest teaching 3.5 Guest disagreement 1.0 Siddhartha pushing back 1.3
speaking balance: gold is Siddhartha, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 16:05 Software companies as daylight robbers
Raja aggressively derides software companies as daylight robbers who take full credit and value while riding on hardware infrastructure built by others.
Hardest push from Siddhartha ▶ 23:05 Pushback on hardware attractiveness to elite graduatesSiddharth directly refuses Raja's initial rationale, stating he is inconclusive about why top engineering talent would abandon software for semiconductors over the next two decades.
Biggest teaching moment ▶ 55:55 Correcting host on L&T's corporate identityWhen Siddharth questions whether L&T is an Indian company, Raja immediately halts the misconception and clarifies that L&T is a fully Indian enterprise investing heavily in semiconductor design.
Siddhartha holds their own ▶ 37:45 Analysis of the 1991 liberalization and IT services modelSiddharth commands the exchange by articulating how currency devaluation and talent arbitrage enabled Infosys and TCS to dominate IT services, challenging Raja to map out India's equivalent semiconductor thesis.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Siddhartha as informed peer | Guest teaching | Guest disagreement | Siddhartha pushing back | Why |
|---|---|---|---|---|---|---|
| Raja Manickam's Career Journey: From TI to Tessolve and iVP Semi | 1 | 2 | 0 | 0 | Siddharth asks straightforward timeline and biographical questions as Raja details his educational background at IIT Kharagpur and his progression through Texas Instruments, Fairchild, National Semiconductor, and Tessolve. | |
| The Evolution of Semiconductors: From Vertical Integration to Foundries | 2 | 4 | 1 | 0 | Siddharth asks foundational questions regarding the definition of a semiconductor and the origins of Silicon Valley. Raja explains the physical concept of conductivity, doping, and the historical vertical integration of early chipmakers. | |
| Moore's Law, Fabrication Complexity, and Analog Power Semiconductors | 2 | 4 | 1 | 0 | Raja breaks down Moore's Law, wafer fabrication processes, and the differences between advanced node computing chips and analog/power semiconductors. Siddharth prompts for clarifications on fabrication terminology. | |
| Hardware as the Highway: Tech Giants Pivoting to Custom Silicon | 4 | 3 | 3 | 1 | Raja colorfully calls software players daylight robbers who exploit hardware infrastructure. Siddharth contributes by discussing Apple's end-to-end integration and custom Apple Silicon chips. | |
| Enabling Global Electrification: iVP Semi and Energy Efficiency | 2 | 3 | 0 | 0 | Raja outlines iVP Semi's mission in power semiconductors and electrification to drive energy efficiency across appliances like air conditioners. Siddharth listens attentively to the commercial applications. | |
| Why Hardware Is Hot Again: Co-Design, Customization, and MEMS | 5 | 4 | 1 | 3 | Siddharth explicitly states he remains unconvinced about why hardware is hot again over software for top graduates. Raja explains co-design and MEMS integration, which Siddharth synthesizes into a motherboard customization analogy. | |
| The NVIDIA Phenomenon: Graphics Compute, CUDA, and the AI Boom | 3 | 5 | 1 | 2 | Siddharth presses Raja on what CUDA actually does for end users and why giants like Intel or AMD fell behind. Raja educates him on GPU parallel computing, AI algorithm optimization, and CUDA's deep firmware layer. | |
| Market Cycles and Tech Giants: From IBM to NVIDIA's Moat | 6 | 3 | 1 | 1 | Siddharth demonstrates strong context by comparing NVIDIA's market dominance to 1980s IBM and detailing the macroeconomic backdrop of India's 1991 IT services boom under Narayana Murthy to query India's semiconductor prospects. | |
| Overcoming Historical Roadblocks: Capital Mindsets and Policy | 5 | 2 | 1 | 3 | Siddharth pushes Raja on the lack of Indian hardware entrepreneurs and critiques Indian family offices and conglomerates for hoarding ventures rather than funding risky tech innovation. | |
| India Semiconductor Mission: Domestic Fabs and Packaging Units | 4 | 5 | 2 | 3 | Siddharth challenges Raja's assertion that Indians lack a hands-on culture by citing automotive manufacturing and mistakenly questions whether L&T is Indian. Raja firmly corrects him on L&T's history and explains the education gap in practical engineering. | |
| The Fabless Model and Cultivating Hands-On Engineering Skills | 4 | 4 | 1 | 2 | Siddharth repeatedly queries the minimum capital threshold required to launch a domestic fab. Raja distinguishes the fabless design model from foundries and explains why starting with half-billion-dollar analog fabs makes more sense than multibillion-dollar leading-edge fabs. | |
| Strategic Imperatives, Import Bills, and Concluding Thoughts | 5 | 3 | 0 | 1 | Siddharth highlights the national security risks of relying on imported chips for domestic military tanks and aircraft. Raja concurs, referencing India's massive semiconductor import bill and the strategic imperative for indigenization. |
Statements from this episode (22)
Manickam: Hardware requires complex global multi-player coordination unlike software
“It is hard because it involves a lot of multiple players that you will have to interact with versus, you know, software, right? You pretty much, you can create your own software package and then so The engagements that you'll need to have within the ecosystem …”
Manickam: Tessolve has trained over 10,000 engineers for global chip firms
“We created probably more than 10,000 engineers that have gone through our system. And they're all over the world. You know, I'm very proud to say that you will find a Tessolve engineer in almost every chip company in the world.”
Manickam: India has many chip designers but zero domestic chip companies
“We have a lot of chip designers, but we don't have a chip company.”
Manickam: Software companies are 'daylight robbers' profiting off hardware engineers
“So I always think that, you know, the software guys are daylight robbers right now. So the highway has been built by this, you know, hard, you know, hardware people, and they just take advantage of it, right? So they build, you know, applications on top of it.”
Manickam: TSMC is the only game in town for leading-edge nodes
“If you want to go into the highest node possible, you have to, you know, TSM is the only game in town.”
Manickam: Tech industry is shifting back to vertical integration for silicon
“I believe now we are going one full circle, right? So all that outsourcing, you know, has happened, right? Now, you know, they're trying to become vertically integrated again, right? So this system companies want to make their own chips, right?”
Manickam: Application-specific custom silicon is no longer prohibitively expensive
“You can now differentiate your products based on the application and it is not too expensive to do that now.”
Manickam: Nvidia's dominance stems from optimizing hardware for its CUDA software
“The reason why NVIDIA is successful is because they actually had this software, AI software, they call it CUDA, right? So now they had that hardware that runs this CUDA very, very efficiently.”
Manickam: Nvidia would not have survived long-term without the AI boom
“If AI was not there, you know, NVIDIA would not have been able to survive in the long term, right?”
Manickam: Nvidia charges $20,000 to $30,000 per module due to zero competition
“But because they're only game in town, they can charge 20,030 thousand dollars per, per module.”
Manickam: NVIDIA cannot maintain a 90% market share indefinitely
“I always believe that you can't hold like 90% market share in anything. Right. So eventually, yes, you know, I know how quickly it is, you know, anybody's guess.”
Manickam: Bangalore produces more engineering graduates annually than all of Taiwan
“Bangalore will have more than, right? You know, the entire Taiwan put together, right? Or Chennai, for example, right? You know, they have so many engineering colleges and all that.”
Manickam: Southeast Asian chip hubs lack talent depth for fabrication
“Malaysia, Thailand, Philippines, and Singapore and all that, they have a lot of manufacturing and not so much on the fab, but on the backend, we call packaging and so on. But the depth of, you know, engineering requirements, right? Are still not there, right? …”
Manickam: Global semiconductor market will double to $1T in ten years
“The semiconductor market is supposed to double, right? It's about half a trillion right now, right? 506 hundred billion dollars is a semiconductor market. It's supposed to double. Right. It took about 40 years. So, you know, to get to this level in the next 10…”
Ahluwalia: Reliance and Adani build semiconductors internally instead of backing startups
“If you see today, if you talk about semiconductor defense Reliance and Adani's of India, They want to do it all themselves and capture the market share rather than investing in new entrepreneurs”
Manickam: It will take India 20 years to match TSMC's manufacturing
“Now on the manufacturing, you know, if you want to be like a TSMC, it's probably going to take 20 years.”
Manickam: India's Semicon 2.0 budget will target ecosystem and product companies
“The two point O is going to be announced, you know, the next budget. And there's significant amount of money that is coming. Now they may focus on the ecosystem guys, equipment guys there are chip product companies and so on. May not be as much for the manufac…”
Manickam: You cannot build a successful chip company without domestic manufacturing
“I'm a big believer that without manufacturing, you can't build a chip company.”
Manickam: Semiconductor companies with in-house manufacturing are currently struggling
“If you look at Broadcom, Qualcomm, they're all, you know, fabulous chip. They don't have any manufacturing. The guys who have manufacturing are struggling.”
Manickam: India is not yet ready for a $5B advanced wafer fab
“But I don't think our country is ready for even five billion dollar fab because that's, you know, very high end, you know, nodes and all, you have to walk before we can run.”