The Twenty Minute VCThe Impact of H1B Visas on Startups in the US & NVIDIA Invests $100BN Into OpenAI
At a glance
WHAT IT’S REALLY ABOUT
AI mega-financing, venture concentration, IPO mechanics, and H-1B headwinds debated
- The panel frames NVIDIA’s reported $100B OpenAI investment as a ‘let’s find out’ experiment that extends the scaling-laws bet by removing near-term capital constraints, while warning the economics may rely on heroic assumptions.
- They argue AI’s current boom is disproportionately a CapEx arms race among a handful of buyers, creating extreme revenue concentration across GPUs and adjacent vendors (like data labeling) because the same few companies are spending aggressively.
- Venture appears more concentrated because an ‘ultra late-stage private-public’ layer has grown on top of traditional VC, driving headlines like ‘75% of dollars to 19 companies’ without eliminating the underlying seed/Series A engine.
- Using Navan’s S-1 as a case study, they explore why companies may IPO before profitability, how comps and ‘not being last out’ matter, and how lockups/secondaries make liquidity slower and messier than press coverage suggests.
- They expect proposed H-1B cost increases to be directionally negative for startups but partially offset by workarounds (e.g., O-1 visas) and by large companies’ ability to simply pay, while lamenting the lack of rational skills-based immigration policy.
IDEAS WORTH REMEMBERING
5 ideasThe NVIDIA–OpenAI deal is a stress test for AI scaling economics.
The group’s core read is that abundant capital and preferential compute access let OpenAI push the scaling-laws thesis until reality forces a stop—creating a clearer (and potentially abrupt) outcome than if funding were rationed earlier.
AI’s headline boom is more about CapEx acceleration than proven revenue maturity.
They separate real app adoption from a spending cycle where a few decision-makers are comfortable investing hundreds of billions ahead of today’s AI revenue base, making suppliers and ‘co-attached’ services boom disproportionately.
Extreme customer concentration is a structural risk masked by current competitive urgency.
NVIDIA’s valuation is unusually dependent on a tiny set of buyers, but the near-term ‘good news’ is those buyers appear committed to spending heavily to win, reducing short-term demand risk while increasing cycle/overbuild risk later.
OpenAI’s advantage is not just money; it’s momentum and access—while monopoly scrutiny lurks.
They note consumer chatbot share can look ‘Standard Oil-like,’ yet current pricing/subsidy means it’s not a classic excess-profit monopoly; still, both NVIDIA and OpenAI must posture carefully to avoid regulatory backlash.
Venture ‘concentration’ reflects reporting categories, not the death of early-stage VC.
Rory argues the base early-stage market remains similar in volume, but an additional private-public late-stage investing layer—naturally more concentrated—now gets labeled ‘VC,’ skewing statistics.
WORDS WORTH SAVING
5 quotesThis is an epic monopoly like we've never seen. Think how much ChatGPT already dominates our lives. It's the Standard Oil of tech.
— Jason Lemkin
You've got this company, what, $4.5T with only six customers. That's bad news, but the good news is all six of them are determined to spend themselves into oblivion to win the prize. It's a fascinating game.
— Rory O’Driscoll
I have no cash whatsoever. I remember feeling... I literally did not have enough cash to fix the roof on my house. I mean, I had enough stock, right? But nothing was more fun than selling my stock at a 70% loss... to fix that roof.
— Jason Lemkin
Founder-friendly has become bullshit, right? And, uh, but it's table stakes. It's table stakes to get into the deal.
— Jason Lemkin
There's no diligence, Harry. There's none.
— Jason Lemkin
High quality AI-generated summary created from speaker-labeled transcript.
