I feel like we’re having the wrong discourse about Mira Murati’s monster funding round.

I’m sure you, like me, have read numerous engagement-bait posts (likely AI-generated) about Mira Murati’s monster funding round. Seriously, kudos on that achievement.

The narratives we’re celebrating:

  • Female founders - There aren’t enough, and this should exemplify women’s power in tech. Having worked with amazing female leaders, this resonates.
  • Pitch deck magic - With just a deck and stellar track record, she executed a massive raise. “So you can too!” But this isn’t novel in the Valley - investing in proven people with clear vision is Investment Thesis 101.
  • Small team, big funding - Huge success possible in the AI age with minimal headcount. I’m pro-AI as a force multiplier, but I firmly believe the people make great businesses.

But we’re glossing over the critical question.

These amazing people - fully capable of achieving their vision - believe they need $2 billion just to get started. That’s staggering.

Yes, brilliant AI minds are expensive… but the core team will likely get hefty equity packages for retention, not cash burn.

$2 billion would buy a LOT of Super Bowl ads, but I doubt that’s the plan.

$2 billion DOES buy a lot of compute.

Pause.

You need $2 billion of compute to launch the product?!

The discourse we should be having

This should be our actual discourse - the sheer compute requirements this AI wave demands. It’s why so many next-gen AI companies run deep in the red. VC dollars aren’t burning on growth - they’re burning on compute. How long is this model sustainable?

We’ve heard about immense utility requirements for all this processing power. I appreciate we need to invest, but is it socially responsible to have 100+ different LLMs (stat sourced from Gemini) when 25 would still represent a competitive landscape with much lower environmental impact?

Colleagues have been working on AGI. Without commenting on their approach, I’ve always been intrigued by one of their USPs:

Requiring much less compute to achieve output.

That seems like a massive solution differentiator.

What we’re not talking about

I’d love to believe we’re having discourse on AI sustainability - how it intersects with reasonable, fundable compute and solving for appropriate balance between societal innovation and environmental impact.

But we’re not.

Until the money train dries up (AND IT WILL), we’ll keep talking about how you too can raise billions with a pitch deck and smart people. After all, my AI tells me a new AI startup launches every 10 seconds.

The pitch-deck story is the comfortable one. The compute-bill story is the one that actually moves the industry. Happy to talk through which one your strategy assumes.