Starting soon, an investment grade company will be able to deploy hundreds of millions of dollars very quickly, to do things that in theory, a startup could do. I would like to live in the world in which those 2 people get "unlimited AI tokens" to pursue their own thing. Instead, I think we're heading to a world where that startup will need to pay a large prepayment, which they will not be able to raise from investors. The way to fix this is by unlocking lending against collateral, rather than lending against credit quality of the offtakers. We must make this work like cars or houses, where a wide variety of folks can get a loan, rather than like semiconductors, where starting a new fab is a geopolitically difficult feat to accomplish.

If you’ve noticed, Chinese models have been released less frequent lately. And when a new model does come out, it turns to be bad as always. There are several possible reasons for this: They’ve been training their models by distilling OpenAI’s models, but OpenAI itself is falling behind because of Opus. Meanwhile, they can no longer distill Anthropic’s models as easily as they used to. All US labs may have achieved RSI , it will be extremely difficult for China to catch up. Distilling models has become nearly impossible. Distillation attacks now get detected within just a few hours of starting

The benefits of abundant intelligence will accrue far outside of San Fransisco. Spent the last year writing about capital, sovereignty, and the coming reordering of economic power. THE COMMANDING HEIGHTS

Just spoke to the Head of AI at a huge company. The #1# thing they're struggling with is... what do I do with the AI laggards at our company? Like most companies, they’ve found that a Super Pareto Principle is true within their org. 10% of their employees are responsible for 90% of the uplift in productivity/token usage from software engineering to non-technical knowledge work. They’ve spent a lot of time & resources on AI enablement from LLM foundations to function-specific trainings and skill building. And yet, the power law of productivity increases hasn’t budged. So now they’re left with a few sobering thoughts and hard questions: 1) 90% of people are overpaid and 10% are underpaid 2) We’d be better off allocating tokens proportionally to the power law, but how do we do that without killing culture? 3) Significant layoffs are likely warranted and wouldn’t impact overall output, but cultural backlash is a big concern? 4) How do we hire more power law people as a non technology company? 5) Is all of our efforts on training and enablement ACTUALLY worth it? 6) How do we not experience a brain drain where the top 10% all leave because they want to be surrounded by other AI-native 10x’ers?

A lot of ink has been spilled on GPU financing lately. I spent over a decade trading ABS on Wall St, and structured over $2B+ of debt financings as an operator (including at @tryramp). Figured I'd try to make sense of this market, so I went back to a framework I knew well: aircraft ABS and aircraft engine ABS. My take: a GPU neocloud runs on airline economics, and the debt behind it looks a lot like aircraft engine ABS. The parallels: • Perishable inventory. An empty seat after the door closes is gone forever. So is an idle hour on an energized H100 rack. • 85%+ of cost is fixed (servers, 10-year rack leases, 10-14% debt service). One more hour of compute costs pennies of power. • Utilization is everything. Airlines break even around 80% load factor, a levered GPU cluster around 65-75%. The returns live in those last 25 percentage points. • NVIDIA is Boeing and Airbus rolled into one. Powered shells and grid interconnects are the new airport slots. • Wall St copied the aircraft financing playbook: take-or-pay contract plus servers in a bankruptcy-remote SPV, 70-80% LTV. Part 1 of my blog post goes into a lot more detail:

We are nowhere near market optimum for flop efficiency. That NVIDIA GPUs can appreciate in price is a leading indicator that real competition hasn't showed up yet. To the degree you are racing to purchase and hoard compute at long timelines puts you at tremendous risk.

whatever you’re working on, no matter how exciting/ambitious you think it is it will probably be 2-5x less exciting when the next model drops (at least) so you should be working on things which have 100-100x the ambition levels, the potentially batshit crazy impossible ideas

Another day, another round of on-demand GPU rental indices surging across the board and reaching ATHs. Even the A100 index, which recently saw some softness, seems to be inflecting a little.