Weekly Intel - 2026-05-24

Weekly Intel - 2026-05-24

I read a lot this week and the same pattern kept showing up. The big AI players are settling things into place: court cases resolving, acquisitions stacking up, layoffs being repackaged as AI investment. The expansive, anything-goes phase of the AI race looks like it’s ending. The consolidation phase has started.

Elon Musk has lost his lawsuit against Sam Altman and OpenAI Musk lost his case against OpenAI, and not because the jury thought he was wrong on the merits. The statute of limitations defense held, so they never had to decide whether Altman “stole a charity” by converting OpenAI to for-profit. That conversion is now legally safe, and I’d watch for more AI startups taking the same nonprofit-to-commercial path. The big precedent risk just got cleared off the board.

Minnesota becomes first state to ban prediction markets Minnesota outlawed prediction markets entirely (Kalshi, Polymarket, all of them) and the Trump administration sued within hours. So we have a state criminalizing something the federal government is actively defending. I have no idea how this resolves, but if you’re building anything adjacent to betting markets or tokenized event contracts, the map is about to get messy. Watch what other states do over the next few months.

AI & Software

Gemini 3.5 Flash Google’s new model is half the cost of competing frontier models, and Google’s pitching it for long multi-step work like building apps, maintaining code, and prepping financial documents. The interesting bit for me isn’t the model itself. It’s the “Antigravity harness” integration that lets 3.5 Flash coordinate teams of subagents in parallel. Google’s betting that intelligence isn’t the moat anymore, coordination is. I think that’s probably right.

The last six months in LLMs in five minutes Simon Willison did a five-minute lightning talk at PyCon and called out something I’d been feeling but hadn’t named: November 2025 was an inflection point. Five different models claimed the “best LLM” title in a single month, mostly for coding. If the leader changes every couple weeks, you can’t really build a long-term strategy around any one provider’s capabilities. I’d ask the question Willison’s basically asking: can your current AI setup swap providers without rewriting half your stack?

Google changes its search box Google is rebuilding Search to host persistent AI agents that watch the web for you and ping you when something matches what you’re looking for. This isn’t a feature upgrade. It’s Google trying to be the layer between you and every decision you make, all the time, not just when you type something into a box. If you run any kind of website, I’d start asking whether your content is built to be found by people browsing results or by agents filtering them. Those are turning into different problems with different answers.

AI Industry Moves

Anthropic acquires Stainless Anthropic bought Stainless, the company that has quietly built every official Claude SDK since the API launched. Stainless turns API specs into SDKs, CLIs, and MCP server connectors. It’s the plumbing that lets agents talk to things outside the model. With chatbots, model quality was the moat. With agents, what plugs into what matters way more. Anthropic owning that layer means they control how Claude connects to the enterprise systems agents will actually need to use. The more I think about this one, the less it looks like buying a tools vendor and the more it looks like buying integration real estate before everyone else figures out how valuable that is.

Mistral AI acquires Emmi AI Mistral bought Emmi AI, a Linz-based outfit that builds physics-informed AI for industrial simulation (energy, automotive, semiconductors, aerospace). Moving from general-purpose LLMs into industrial engineering is a real shift. Those are high-margin customers with sticky integrations. I think Mistral’s playing for a specific customer here: European industrial companies who don’t want US hyperscaler dependence and need genuine engineering depth, not just chat. Whether they can actually pull that off is the open question for me.

Intuit to lay off over 3k employees to refocus on AI Intuit cut 17% of its workforce, about 3,000 people, and called it an AI refocus. Tech layoffs are over 100,000 this year and most of them are using some version of the same framing. CEO Sasan Goodarzi’s $36.8 million package is untouched. I don’t know whether employees and investors will keep buying this story, or whether Intuit’s AI bet will actually produce new revenue. What I do know is that “we’re investing in AI” has become the polite cover for headcount cuts, and at some point the gap between the story and the results closes one way or the other.

Cybersecurity

GitHub confirms breach of 3,800 repos via malicious VSCode extension One GitHub employee installed a bad VS Code extension. An attacker walked away with 3,800 internal repositories. GitHub has now linked it to the broader TanStack npm supply-chain attack and a malicious version of the Nx Console extension, which is a tool a lot of developers trust. This has been getting worse for a while. The attack surface is the software supply chain, and the way in isn’t a phishing email or a zero-day. It’s a developer productivity tool from an official marketplace. If GitHub’s own engineers can’t safely install something from the marketplace, I’d be asking some hard questions about what visibility your team has into what’s running on developer laptops.

GitHub is investigating unauthorized access to their internal repositories GitHub confirmed it’s investigating unauthorized access to internal repos. That’s about all we know right now. When the platform that hosts most of the world’s source code gets breached internally, the blast radius isn’t just GitHub. Every company whose CI/CD, secrets, and deployment depends on GitHub is in the splash zone. I’m watching for one thing specifically: whether any signed commits, Actions, or internal tooling got tampered with. That’s the part that would matter most.


That’s what I’m watching. What caught your attention this week?

-Eric

Share

Get weekly insights on technology leadership

One idea per issue. No spam. Plus a free guide on measuring AI initiatives when the old metrics don't work.

Or download the free guide directly →