Topic
The costs, scaling, and procurement of computing power—training and inference—that shape AI firms' competitiveness and business models.
An essay on interconnects.ai argues that rising AI-risk expectations at OpenAI and Anthropic follow from watching agents work at scale, and its author says he holds high uncertainty about that reading.
Reality
Sam Altman told Fortune that a listing now would be ill-advised while safety and alignment work is unfinished. More than $180 billion has gone into OpenAI, on Fortune's count, with no scheduled exit.
Reality
Where a robot's inference runs decides what the machine costs to build. SemiAnalysis puts the industry's upfront compute bill in the trillions at a billion robots, paid by manufacturers on every unit before it does any work.
Reality
The 1.7x to 3.6x latency range is set by the baseline systems, not the chip, and the report's own publication date is unsettled. Read it as direction, not evidence.
Perspective Coverage
9 publishersReality
Her P99 keynote sets hardware aside because most teams cannot change it. That leaves the weights and the serving layer, and her own goodput example shows where the seconds in a response actually go.
Reality
OpenAI says the run took 88 hours and cost millions, which tells a small research group more about whether it can compete than the proof itself does, and its wording on user data shows the limit of what any vendor can promise.
Reality
Export controls have governed who holds a chip rather than who logs into it. A draft that could reach AI companies in September would change that. The deals already done show what offshore capacity was priced to earn.
Reality
A new study says AI agents still fail at free-form research. The self-improvement story is the load-bearing beam under a lot of compute spending, and it just got a crack in it.
Reality