Happy Sunday and welcome to Investing in AI. If you haven’t checked out our AI Innovator’s Podcast recently, we have some great new episodes. We also launched our AI Innovator’s Investment Syndicate last year. If you want to apply as a company or angel, please sign up here.
I recently picked up a copy of “The Thinking Machine” by Stephen Witt, and read the book in two days. I could not put it down. It covers the history of NVIDIA from inception to the modern day AI boom, and all the chaos along the way. This is company that almost died multiple times, made Jensen a billionaire in the late 90s, but then lost most of it’s market cap after the web 1.0 bubble and didn’t recover for 14 years.
Two things jumped out at me from the story that are worth sharing. The first is – Wall Street used to hate CUDA. Yes, that’s right, the software package that is now considered a big part of NVIDIA’s moat was panned by wall street for years. When Starboard, an activist investment firm, took a large stake in NVIDIA in 2013, one of the big concerns was that they would force NVIDIA to stop investing in CUDA. Luckily Jensen was able to work something out and Starboard’s activism was minimal. But, it’s a tough lesson for investors that sometimes moats take a long time to build and aren’t obvious in the beginning.
The second point, which is related, is that Jensen used to call CUDA a “zero billion dollar market.” In investor meetings, when asked about the market for CUDA, he would tell them “there is no market for CUDA currently.” Jensen would tell employees it was a zero dollar market with no current demand, but that someday would be a billion dollar market.
The book is filled with stories like this – NVIDIA taking big bets, some of which paid off, some of which didn’t. At times in their history, they had the best graphics cards on the market, and at other times they were surpassed by competitors for a few years.
The lesson, as investors, is that sometimes to be able to take advantage of market shifts, technology shifts, and consumer demand shifts, you have to already be in the game. Overanalyzing the future through the lens of the present is useless, even though that’s what most investors do.
Think about this – NVIDIA built one of the most valuable companies in the world by ignoring all the things we tell people to do in business.
-
CUDA did not “solve a hair on fire problem.”
-
CUDA didn’t help customers with a “job to be done.”
-
Customer development on CUDA would have told you there was no market, and no need.
-
NVIDIA invested in it despite knowing there was currently no demand and it was a zero dollar market.
I point this out because I think investors in most market stages have moved away from backing big visions, in favor of efficient iteration. But I remember meeting a very senior CEO at one of First Round Capital’s entrepreneur days in 2010, when lean startup thinking was getting really hot. He told me “great companies are built on bold vision, not iterating your way to product market fit.”
As you place your best on the AI future, don’t forget the CUDA story.
Thanks for reading.