You Gotta Invest in What You Know

The most important piece of tech media that came out this week was a FANTASTIC interview between Jensen Huang, Founder & CEO of NVIDIA and Brad Gerstner, Founder & CEO of Altimeter. I’d highly encourage anyone to watch the video end to end. It’s jam packed with a ton of insight for both founders and investors alike. Jensen is clearly an athlete operating at peak performance. There is a depth and precision to his conversation, and the best way to describe him is a master. He is a master of entrepreneurship, a master of semiconductors, and a master of the global AI industry. Brad Gerstener is no slouch as well, and there’s a lot of nuggets that technology investors can pick up from him too. There’s an ebb and flow to his questioning, a pliability to how he approaches the conversations that alternates betwen enthusiastic alignment and gentle skepticism. There’s no arguing that these two individuals are at the top of the capitalistic food chain. It’s truly a wonder that literally anyone, has the opportunity to tune in, as these two prime movers discuss the global economy, and the biggest technological trend of the 21st century.

Three things stood out to me from the interview:

1. The idea of market dominance/monopoly opportunity

2. Transitioning the economy to augmenting human intelligence

3. Investing in what you know & speaker credibility

Market Dominance / Monopoly Opportunity

If you spend any amount of time within the tech ecosystem, especially amongst VCs, you will at some point be pointed toward Peter Thiel’s book Zero to One. The central theme of that book is to escape competition, and pursue monopoly opportunities. This is an idea that is simple in theory, but nearly impossible to do practice. It’s also an idea that came up in the discussion between Jensen & Brad. (Lol look at me, referring to them as if we’re boys). There’s a point in the interview where Jensen & Brad are going back and forth about the competitive dynamics at play for NVIDIA. The game on the field is that NVIDIA is facing pressure from players like Google & Amazon, among others, which are building out custom ASICs that can specifically serve as an alternative to NVIDIA GPUs. Jensen’s POV essentially boiled down to (and i’m summarizing here), “hey these guys are coming into the game at the 11th hour. they are formidable opponents for sure, but they are entering the market after I’ve spent 30 years defining it, and alongside those 30 years of market definition, I’ve built the relationships, and connections, and connectivity, and trust with buyers, suppliers, governments, you name it - to pretty much be the only game in town that can deliver on the current demands of the current marker environment.”

The specific quote that Jensen had was “you want to own 100% of a small industry,” and I found this to be so enlightening because it goes against the grain of all conventional VC dogma. VCs breathlessly focus on TAM, TAM, TAM, TAM and while it is INCREDIBLY IMPORTANT, if you over rotate on conventional views of TAM - you miss out on opportunities like NVIDIA. You miss out on opportunities like Scale AI, you miss out on opportunities like YouTube. I acknowledge the survivorship bias in my thinking, but backwards looking all three of these businesses did not have well established multi-billion/multi-trillion dollar markets to sell into. These business were ahead of the market curve, and extremely well positioned to take advantage of the twin forces of multi-decade compounding and exponential growth.

Augmenting Human Intelligence

If I had to summarize the entire interview in one line it would be this: “Jensen Huang believes that every knowledge worker in the world, will have an AI assistant that makes them more productive and he’s actively taking orders from the manufacturers of intelligence (i.e the model labs) to pull forward that future.” This is something I fundamentally agree with, and wrote about back in 2024, in my article Service is the new SaaS. If you make your living sitting in front of a computer each day, you are going to have a computer to help you do your work faster. Software is going to be running all the time, 24/7, and software switching from 0% marginal cost, deterministic instructions to probabilistic systems with a marginal costs set by the price of energy.

What was most interesting about this part of the discussion was Jensen’s assertion that “we sit at the end of the supply chain,” meaning NVIDIA builds to the demand specified by their partners. I don’t have perfect insight into how the company runs, but it’s a bit like saying we build to order, not we build to forecast. The way I think about it is, he’s saying a customer, let’s say OpenAI tells NVIDIA, we have plans to build out data centers that will need 400K GPUs, and we want those data centers to go live in 3 years. NVIDIA says okay, we’ll spin up our supply chain to make sure we can deliver those 400K GPUs over the time frame you need them. And that’s the way we’ll run the business rather than saying, we think we can sell 400K GPUs in the open market, and we’ll find people to buy them. The subtle difference in a demand pull, rather than a demand push leads me to think about the world in this way:

• If NVIDIA sits at the end of the supply chain, the front of the supply chain are the applications that augment human intelligence

• These are both the foundational models themselves, but also pure application companies like Harvey AI, Rogo, and OpenEvidence

• These end applications are the canaries in the coal mind of intelligence penetration

• How much of a lawyer, or a doctor, or a banker’s work is run through some sort of intelligent workflow today? My guess is it’s somewhere between 1 - 5%

• As these end applications grow their intelligence penetration from 1- 5% to 15 - 20%, that sends additional demand signals to the foundation models to A) build out more data centers to handle this continious workload, and B) optimize incremental data centers to be more efficient for their own profitability

• As I wrote about last week, I believe the AI has a “last mile” problem because historical software companies aren’t purpose built for AI, nor do they necessarily always have the talent to build these new AI native solutions

• I’m at maximum conviction in my belief that the application layer is going to mint companies that generate, in aggregate, $100s of billions of dollars worth of revenue and if you believe in that world, then, based off today’s information, NVIDIA is in an incredible position to be the catcher’s mitt of infrastructure to run this continious intelligence

Speaker Credibility

My final takeaway from the conversation was speaker credibility. We live in such an incredibly noisy media environment. There’s a lot of bullshit, and a lot of people who don’t know what they are talking about. I’m not afraid to admit I’m not an expert in global semiconductor supply chains, or the debate around open versus closed source software, or the scaling laws. But I am very confident in a few things. I can recognize expertise when I see it (Jensen is an expert), I can sniff out bullshit when people aren’t being honest about their incentives, and I after 4 years I’m becoming an expert in some aspects of the technology industry and VC. My favorite part of the entire interview comes pretty early on, where Jensen, when pushed on the particulars of his recently announced $100bn investment into OpenAI, he says, “you gotta invest in what you know.” It’s again, a simple in theory, more difficult in practice idea and it made me think about the credibility of any speaker. Jensen Huang has 100x more visibility into the demand pull of the AI industry. Is he incentivized to fufill that market demand, and be a global advocate of the benefits of AI? Yes. Is he disincentivized, to overpromise and over commit? Also, YES.

Unfortunately in today’s environment, there’s a fine line between telling a compelling story and lying. There’s a Mark Twain quote that’s stood the test of time which is, “never let the facts get in the way of a good story.” I think a lot of people in business and politics, especially successful people, internalize and practice this belief at some point or another. Good, bad, or ugly, it’s just a fact of life. But what is also a fact of life, and certainly a fact in markets is that chickens come home to roost. If extraordinary stories lead to ordinary results, companies atrophy and die. If extraordinary stories lead to extraordinary results, companies compound and I do think management, CEOs, like Jensen Huang are the earliest signal, of success or failure. No one can be 100% credible, 100% of the time. The world is too complex and there are too many variables shifting all the time, but when you listen to entrepreneurs like Jensen, who have been consistent, who have been pragmatic, who possess an agility an ability, a dexterity and determintion or as Jensen put it in his own words “will meets skill” you’re likely going to be in a pretty good position, for a pretty long time provided they are still in their seat. So goes the CEO, so goes the ship, and Jensen Huang is simply one of the best in the game right now.

All in all, this interview was a fascinating conversation and I would encourage anyone to tune in to the full interview here. Peace! And talk to you all next week.