An LNG Exporter Is Building 10 GW of Compute Because It Already Owns the Fuel
Venture Global will build up to 10 GW of compute behind the meter, making off-grid data center power the way around an interconnection queue.
India Rated AI Agents Critical. Its Own Fix List Is a Secrets Audit
India rated AI agent security critical after agents read personal records. Its own root-cause list is a leaked-credential audit.
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.
vLLM Let Coding Agents Write Its Kernels. Review Became the Hard Part.
vLLM kernel optimization by coding agents cut a Kimi K3 layer from 23 kernels to 13. The review process is now the bottleneck.
Five Constraints Moved This Week, and Every One Landed on Your Invoice
AI infrastructure costs moved on finance, grid, licensing and tokens this week, and every move landed on the operator's invoice.
ServiceNow Wants Every AI Agent to Carry a Passport, and Revoke It at Machine Speed
AI Passport is ServiceNow's signed credential that agents and MCP servers present at runtime, enforced by NVIDIA OpenShell and Sentry on the DPU.
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.
Broadcom Pulled the VDDK. Nutanix Rewrote the Migration Path Around It
Broadcom pulled the public VMware VDDK download, so Nutanix Move 6.3.3 adds DirectPath to migrate ESXi virtual machines without it.
Disaggregated Inference Added a Latency Tax. More Bandwidth Was Not the Fix.
A push-based connector in vLLM fixes a latency tax in disaggregated inference, where KV transfer setup stalls the decode worker.