A quiet start to the month for Serenity, the stock-picker whose reconstructed views this site tracks, but the one item he did raise pointed straight at his core obsession: the widening gap between AI compute demand and the industry's ability to build it.

His prompt was a fresh headline that Anthropic, the AI lab behind the Claude models, has signed a roughly $35 billion compute agreement with Lambda, a GPU cloud provider backed by NVIDIA (NVDA). Serenity's takeaway was less about the two companies involved than about the pattern: frontier model developers are locking up capacity faster than the ecosystem can physically deliver it.

That framing is central to how Serenity invests. NVDA sits at the middle of his entire framework — not as a holding (he has said repeatedly he does not own it, viewing it as a crowded mega-cap with lower upside than the names further upstream) but as a demand signal. He treats NVIDIA's roadmap and the scramble for its GPUs as a telegraph for where scarcity lands next, then hunts the smaller "chokepoint" suppliers — the critical upstream vendors with few substitutes — that quietly feed the buildout. A $35 billion deal from a single lab, in his reading, is another data point that the demand side keeps escalating.

He also drew attention to Hut 8 (HUT), a former bitcoin miner that has been repositioning its power-heavy sites toward data-center and AI compute hosting. Serenity paired HUT with the Lambda arrangement to illustrate the plumbing behind these deals: labs need GPUs, but they also need the data-center shells, the interconnect and — crucially in his world — the electricity to run them. Companies that already control powered land and grid connections become scarce infrastructure once the AI capacity race accelerates, which is the "boring sector about to re-rate" logic underpinning his AI-power theme.

The open question he posed was which neocloud or colocation player gets locked up next. Serenity tends to treat AI cloud capacity as a long-short theme — favoring names he sees as undervalued on a sum-of-the-parts basis while flagging over-levered operators as structurally weak — so a wave of mega-deals like this one matters to him both as confirmation of demand and as a sorting mechanism between the winners and the stretched.

Nothing in today's note amounted to a new position or a changed thesis. It read more as Serenity keeping score: watching the compute deals pile up as evidence that the bottleneck-driven scarcity he bets on — in lasers, substrates, memory, packaging and power — is only tightening. For readers, the signal is directional rather than actionable, a reminder that his front-running approach leans on exactly these demand markers to decide where the next squeeze will show up.

*This is derived commentary based on Serenity's public posts, not investment advice, and does not reflect his exact words.*