Friedberg draws an analogy to 'speed doubler' internet companies from 2003 — transitory arbitrage businesses that were disrupted when broadband arrived — warning that hyperscalers spending $80B+ in CapEx each may commoditize what CoreWeave currently offers, compressing demand and pricing.
I worry a little bit about a business like this where there's 4 or 5 companies that are each doing $80 billion of CapEx this year to create infrastructure that effectively starts to replace what these guys are effectively offering out as a” ⚑
Chamath is impressed by CoreWeave's technical differentiation (bare-metal, no hypervisor) and revenue trajectory, but flags that the key investment risk is whether the assumed useful life of NVIDIA GPUs proves correct — if it's shorter than modeled, the business is deeply underwater.
The bet is they've built an amount of headway. They're going to continue to do this good technical engineering. But the other side of it is, is the useful life right? Is the technology curve right?… we thought the useful life was 10 years,” ⚑