NVIDIA

Jensen Huang

Spent thirty years building for a market that did not exist yet.

The longest-horizon bet any of us have watched pay off, and it nearly died twice on the way.

1. Build the thing the future needs and wait.

   CUDA shipped in 2006 for general-purpose GPU computing. The workloads that made it essential arrived years later. That gap is the story.

2. Survive the near-death moments.

   NVIDIA came close to failing more than once in the nineties and 2000s. Nothing compounds if the company stops existing.

3. Pay for the platform before the demand.

   Investing in software nobody was asking for looked like waste for a decade. It became the moat.

4. Founder tenure is an advantage.

   Running one company for thirty-plus years lets you make bets no professional manager on a four-year cycle could justify.

5. Flat beats layered.

   He is known for an unusually flat structure and a very large number of direct reports, which keeps him close to the actual information.

6. Explain the technology yourself.

   He does the explaining, at length, in public. A founder who can teach the market is worth a marketing department.

7. The adjacent market is the real prize.

   Graphics cards for gamers funded the compute platform. Cash flow from today buys the option on tomorrow.

8. Discomfort is the working condition.

   He talks about pain and suffering as ordinary parts of building, not obstacles to be removed. It is a more durable stance than optimism.

9. Hire for capability, keep people a long time.

   Low churn in senior ranks means institutional memory, which is what lets a thirty-year bet stay coherent.

10. Do not confuse the payoff with the plan.

    The outcome looks visionary in hindsight. Read the interviews from 2010 and it looks like conviction under sustained doubt, which is what it was.