Tag
Nscale's S-1 discloses $103.4 billion in contract value from Microsoft and Anthropic. Only $2.6 billion of it was active revenue as of August. Here's how to read the gap.
ERCOT's queue holds 474 GW of data center interconnection requests, about five times peak demand. Most of it is speculative filing, not real growth, and utilities are now pricing that in.
Two 2026 reports on Nvidia's inference market share reached opposite conclusions. Both are accurate — they're counting two different chip markets that happen to share a name.
Microsoft shipped seven MAI models five weeks after a contract amendment capped what OpenAI owes it at $38 billion through 2030. Read the two events together and the launch looks like a hedge, not a roadmap milestone.
Moody's says hyperscalers carry $662 billion in data-center leases that haven't hit their balance sheets. Add stretched GPU depreciation, and the capex number everyone quotes is the smallest one in play.
Vendor benchmarks leave out the two cost items that usually flip the self-hosting decision: engineering overhead and the model-update cycle. Here is the honest break-even analysis.
The 80% savings claim for self-hosted LLMs is arithmetically correct on a fully-loaded GPU. Here is what the calculation looks like when you count correctly.
Practical guides, product updates, and compliance notes — straight to your inbox. No fluff.
Newsletter is opening soon. We'll switch this on once we've got our first issue ready.