Infrastructure
A 14 Percent Crash Will Not Cut Your Memory Bill
Samsung fell 13.4 percent and SK Hynix 14.7 percent on a Chinese lithography tool. That tool makes commodity DRAM, not HBM. What it changes in your memory contract, and when.
Infrastructure
Infrastructure
Copilot Is Served Before Azure Customers
Microsoft's finance chief told analysts Azure would have grown past 40 percent if new GPUs had gone to customers rather than first-party Copilot. Capacity is now a physical limit, not a commercial one.
Infrastructure
OpenAI Paid You Back in the Thing That Broke
Four incidents on 27 July alone, on OpenAI's own status page. Compensation for the weekend outage was a usage reset, which is credit on the service that failed.
Infrastructure
Amazon Filed For 5,105 Satellites It Cannot Yet Use
Amazon asked the FCC to fly 5,105 direct-to-device satellites from 2028, using spectrum it does not own yet. What that timing means for your continuity plan.
Infrastructure
Cadence And Synopsys Tuned Their Tools For Nvidia
Nvidia says its own Vera CPU runs Cadence and Synopsys verification up to 1.5x faster. The number is Nvidia's, and the design farm question is now open.
Infrastructure
Nvidia May Guarantee 250 Billion of OpenAI's Rent
OpenAI holds no investment-grade rating, so Nvidia is reportedly guaranteeing the lease on a 10 gigawatt Ohio campus. A chipmaker's credit is now a line in your compute budget.
Infrastructure
Anthropic Asked a Memory Maker for Its Own Chips
SK Group's chairman said on 25 July that Anthropic approached SK Hynix for supplies to build its own semiconductors. That adds a claimant to the queue you are in.
Infrastructure
Multi-Homing Does Not Mean You Choose the Path
Cloudflare measured what other networks do to a BGP attribute it sets. About 70% of paths came back rewritten, and the biggest networks are the ones doing it.
Infrastructure
AMD Handed Half of Inference to a Rival Chip
On July 23 AMD and Cerebras built one inference pipeline from two rival chips: AMD Helios handles the prompt, the Cerebras wafer writes the answer. The 5x efficiency figure is a model, not a benchmark. Here is what disaggregated inference changes for your compute budget.
Infrastructure
256 Cores on One Chip Changes Your Rack Math
At its Advancing AI event on July 22 AMD launched EPYC Venice, a 256-core server chip on TSMC 2nm claiming a 70 percent efficiency gain. Here is what that does to your rack count, your power bill and your negotiating position.
Infrastructure
Nvidia Stopped Renting Its CPU Core From Arm
Nvidia's new Vera server CPU runs Olympus, its first in-house core, not licensed Arm cores. The three percent benchmark win is not the story - the lock-in is.
Infrastructure
Anthropic's Fix for Claude Outages Arrives in 2027
AMD will invest up to $5bn in Anthropic and supply 2 gigawatts of MI450 chips. The relief for Claude's rate limits is real, but the first gigawatt is dated to 2027, and your chip supplier now owns a stake in its own customer.
Page 4 / 8
The Operator Brief
One considered note on infrastructure, governance, and measurement, most mornings. No theory.