Skip to content
MustHave.ai
MustHave Hoodie Seal
THE MUSTHAVE NEWSROOM

AI news

What’s new. What matters. What do you think?

Wednesday, August 12, 2026

Aug 12

NVIDIA NeMo Switchyard lets Codex use local models. Test the fallback first

NVIDIA NeMo Switchyard routes Codex and Claude Code across local and cloud models. Here is what to test before the pre-alpha proxy touches a real repo.

Go AI-assisted software engineering: we tested Google’s verification loop

Google says Go gives coding agents a tighter verification loop. We ran a disclosed micro-test against Python and TypeScript to see what the defaults catch.

Riot confirms a 191 MW AI lease. The customer is still unnamed

Riot confirms a 20-year, 191 MW AI data-center lease worth about $9.1B. Reports identify Anthropic; Riot’s official filing still leaves the tenant unnamed.

CoreWeave doubled quarterly revenue. Interest expense still reached $640M

CoreWeave Q2 2026 revenue hit $2.575B, while net interest expense reached $640M. Here is what buyers should read beyond the growth headline.

CME launches AI compute futures. Here is what the hedge misses

CME plans H100 and B200 compute futures for October 5. Here is what the rental-price hedge can cover, what it cannot, and why launch is not final.

OpenAI says ChatGPT Ads hit a $1B run rate. Targeting is the bigger story

OpenAI says ChatGPT Ads reached a $1B annualized run rate. The sharper questions concern conversation context, custom audiences and measurement.

Copilot Memory now follows JetBrains work across sessions. Check what survives

GitHub Copilot Memory and Ollama support are in JetBrains 1.15. Here is how memory scope, expiration, admin controls, and local-model claims work.

MAI-Code-1.1-Flash costs a quarter request in Copilot. The old model is leaving

MAI-Code-1.1-Flash brings image input and a 0.25× Copilot request multiplier. Teams have until September 10 to retire the old model.

Google put sign-language AI into Gboard. The server boundary still matters

Google's SL2T turns ASL into English text on Pixel 11. Here is what runs on device, what reaches Google's servers, and where errors still matter.

Tuesday, August 11, 2026

Aug 11

Nemotron 3.5 Lightning is a 66 GB model with 3B active. Do not confuse compute with memory

NVIDIA's Nemotron 3.5 Lightning activates 3B of 30B parameters, but the BF16 weights are about 66 GB and one-million-token context needs serious hardware.

Spotify’s AI Persona badge labels the artist, not the song

Spotify will label some AI-generated artist identities and remove them from default recommendations. The badge does not explain how every song was made.

Claude Watermarks Expand to Older Models; Detection Remains in Private Preview

Anthropic expands text watermark support to older Claude models. Sonnet 4.6 sampling settings can affect the signal, while detector access remains restricted.