NVIDIA (Founding)
NVIDIA (Founding)
USA, Founded 1993
Society
8/10
Customer
4/10
Self
4/10
1 above the survival line
Society
8/10
Customer
4/10
Self
4/10
1 above the survival line
Three guys in the corner of a Denny's. A bet placed in an era when 3D graphics was dismissed as 'a toy for hobbyists.'
12 criteria
32/60
3 axes
16/30
NCG
48/100
Context
April 1993. At a Denny's in East San Jose, three founders met: Jensen Huang (30, ex-LSI Logic and AMD), Curtis Priem (designer of Sun's GX graphics chip), and Chris Malachowsky (ex-Sun). Started with $40K of self-funded capital. At the time, PC 3D graphics was treated as 'a toy for gaming hobbyists,' and the industry consensus was that the CPU was enough. Their first product, NV1 (1995), backed the wrong standard and was a commercial failure. The company had not yet been saved by Riva 128 (1997).
Teles' Insight
The starting point of the company that would become the world's most valuable in AI silicon, 30 years later, was three people in the corner of a Denny's in East San Jose. Jensen Huang at 30, $40K of personal money, zero product, and an industry that laughed at 3D graphics as 'a toy for hobbyists.' Their first product, NV1, picked the wrong standard and flopped. A normal founder would have quit in 1995. Instead, over 30 years, they walked the path: Riva 128 (1997) → GeForce (1999) → CUDA (2006) → deep learning (2012) → ChatGPT (2022) — proving the thesis that 'graphics processing is the foundation of all computing.' A rare case of a company that started below the survival line at 32 points and climbed all the way to S-tier.
Key to survival
When the entire industry called 3D graphics 'a hobbyist toy,' they held onto a counter-thesis to the 'CPU is enough' consensus, and refused to quit even after NV1's 1995 failure. Starting from $40K and grinding the same thesis for 30 years, that persistence bloomed — at the arrival of the AI era — into 20 years of software-compatibility moat.
12 Criteria (32/60)
WeaknessesC Solution, E Traction, F Moat, G Revenue, H Financial, J Exit
PC 3D graphics was nascent and niche. The gamer population was small. SAM was tiny.
3D math on the CPU was slow. Gamers felt it, but it wasn't a 'business-critical pain.'
A dedicated GPU to accelerate 3D math. NV1 backed the wrong standard, so the product was unproven.
Jensen Huang (LSI / AMD) + Curtis Priem (Sun GX) + Chris Malachowsky (Sun). Complementary skill sets — strong for the time.
Zero product. Still writing code. Only $40K of personal capital.
No barrier. Competitors everywhere — S3, ATI, Matrox, 3dfx, SGI. Patents and accumulated know-how were yet to come.
Planned chip-sales model. Unit economics unproven.
$40K self-funded. Three years later they would raise $20M from Sequoia and others — but at founding, a textbook seed-stage startup.
Bootstrapped early. Pre-Sequoia, the company was too small to evaluate capital efficiency.
No exit design. IPO would not come until 1999, six years later.
Just before Windows 95. PC adoption and 3D gaming demand were both rising. Maybe two or three years early, but the inflection point was broadly right.
'3D graphics will become the foundation of all computing' — a counter-thesis to the industry consensus ('the CPU is enough,' '3D is for hobbyists'). Competitors at the time believed the opposite.
