@viggy_krishnan@matt_slotnick@zephyr_z9 “We estimate Anthropic run-rate revenue as of 6/14 at ~$62bn, up from an estimated ~$54bn at the end of May-26 and ~$47bn earlier in May. Sequential ARR added accelerated in Jun-26 MTD driven by Cloud API and OpenRouter, while Consumer spend growth continued to slow”
@DMTCapital@PythiaR@TShirtnJeans2@S_curvecap GOOG + MTK because AVGO has a direct procurement approach, purchasing HBM straight from memory suppliers to consolidate/exert its own significant buying power. With MTK, GOOG place HBM orders directly with suppliers, and provide the components to MTK on a consignment basis.
@viggy_krishnan@evrgn11112231 But given it’s highly tied to OAI needs, the question is if MSFT can abstract the token to use for other workloads beyond OAI… otherwise will have very little value for the entirety of Azure customers. Right?
@ContrarianCurse@TwannsWorld@P_Remarks Do you think they would be growing 30%+ without accelerators doing the recommendation engines? This 6% is only one part of the story.
@teslayoda Tesla has almost 300k H100 equivalent… rent this shit and can generate about $10B annual revenue. More than double Tesla EBIT! I’m quite sure they don’t need all those Nvidia chips to train FSD and Optimus
@ContrarianCurse I thought Ant clear focus on enterprise + OAI fumble with Sora / agentic commerce and a pivot to enterprise would be a clearing event for internet stocks in general… but clearly not
@FredaDuan Is it fair to compare this deal with CoreWeave on-demand pricing? Here we’re talking more than a handful of GPUs + you have the 90-day notice period… makes the premium to spot price even more puzzling
@JaredSleeper Is it really all consumer (eg Claude chatbot) for Anthropic? I suspect there’s Claude Code tokens usage for small developers in the mix. Remarkable anyway!