The history books will be very kind to Jensen Huang.
Trillions in economic value created...all downstream from his visionary leadership decisions.
One decision was on CUDA.
Nvidia started funding it in ~2006 and kept funding it for years while it dragged margins. There was no AI market to justify the spend which led to huge skepticism on the Street.
His foresight will prove to be legendary.
The second was on Mellanox.
Paying $6.9B for a networking company in 2019 was viewed as a very odd move for Nvidia. In fact, I'd say that it's the deal that let Nvidia sell the whole rack instead of the chip which is where much of Nvidia's pricing power comes from today. Very few CEOs get a call like that right.
Absolute legend.
The most mind blowing stat we've seen so far is this:
"As of June 30, 2026, we had raised an aggregate of $141.1 billion of capital and had cash and cash equivalents of $112.4 billion and short-term investments of $19.3 billion."
That means to date, they have burned 9.7b, they are growing at a rapid pace, and are profitable on an operating income basis as of the June quarter! This is likely the most efficient scaling of a business in the history of capitalism.
https://t.co/bWYbqKOMDN
@masayoshisson@3F_Research 6 years seems conservative tbh.
I think with disaggregated inference you will see Blackwells have useful lives over 10 years.
Will be able to put the latest and greatest SRAM accelerator in front of a B200 in 5 years and make it sing.
@socialcapital@grok Of the post-training/RL share, what fraction is rollout generation vs reward/verifier vs gradient step? If its mostly rollouts (forward only generation, no backward pass), then isnt that compute technically just inference?