Your 2029 AI Capacity Plan Is About Three Times Too Big
A new AI data center capacity forecast expects hyperscalers to build 5.86 GW in 2029 against 18 GW announced.
Amazon Wants Decart, and AI Chip Portability Is the Prize
Amazon is in talks to buy Decart for about $7 billion. The prize is AI chip portability, software that lets a model leave Nvidia.
Modal’s $1 Billion Run Rate Rests on a Runtime It Wrote Itself
Modal crossed a $1 billion run rate selling inference cloud capacity. A third of it comes from sandboxes, not GPU hours.
Perplexity’s Agent Sandbox Held the VM Wall. The Network Wall Is Where It Broke.
Perplexity open-sourced its agent sandbox, SPACE, with a published escape test. Nine models ran 108 attempts. The VM held and the network policy broke.
QumulusAI Is Billing 7 MW of AI Capacity Ten Weeks After Signing
QumulusAI, an AI neocloud, is billing on 7 megawatts of Blackwell capacity ten weeks after signing, with $523.4M in signed contracts.
Intel Is Selling Inference by Memory, and Its Own Card Holds 160 GB
Intel GPU D-series promotes 480 GB of LPDDR5X for inference. The card Intel itself sells carries 160 GB, and the gap decides the purchase.
The CPU’s Memory Is the Next Line on the AI Power Bill
AMD EPYC Verano is the first EPYC to carry LPDDR5X in SOCAMM2 modules, and server memory is now a rack power decision.
Nvidia Is Paying Researchers to Attack the Agent Sandbox It Built
Nvidia pledged $250M in compute for UK agent safety research. The money funds the adversarial testing its own sandbox still needs.
PJM Wants to Review a Ten-Gigawatt AI Campus as One Project
PJM filed to review a ten-gigawatt Ohio AI campus as one integrated project, a new path around the data center interconnection queue.
The AI Buildout Just Met Its Cost of Capital
AI infrastructure spending now runs about five trillion dollars ahead of the AI revenue that pays for it. On Wednesday the bond market priced the gap.