AI buying just started looking less like software and more like hiring a firm

Listen to the “podcast” style version of this issue below!

Generated with Google’s NotebookLM.

Until now, buying AI felt like buying software.

This week, it started feeling like building infrastructure.

Capacity got real. Agents arrived by default. The labs themselves moved into services delivery. The ROI math was quietly rewritten.

On the same day, Anthropic and OpenAI announced separate joint ventures to deploy AI services inside mid-market companies.

Anthropic announced a deal with SpaceX to use all compute at the Colossus 1 data center in Memphis. 300+ megawatts. 220,000+ NVIDIA GPUs.

Earlier in the day the Claude Code team released major updates that make the Claude agent ecosystem more robust and consistent.

Anthropic's venture: $1.5B valuation, backed by Blackstone, Hellman & Friedman, and Goldman Sachs. The model is embedding AI engineers with client companies to deploy Claude into core operations.

OpenAI's: a $10B target valuation, called The Development Company or "DeployCo." Backers include TPG, Brookfield, Bain Capital, and 16 others.

The labs are not staying in the API business. The implementation work that Accenture and Deloitte have done for two decades is now what Anthropic and OpenAI are selling directly..

The same week, Anthropic also announced a deal with SpaceX to use all compute at the Colossus 1 data center in Memphis, and agent tooling that turns those raw capacity gains into something you can actually deploy.

300+ megawatts. 220,000+ NVIDIA GPUs.

Capacity for Claude Code doubled across paid tiers. Opus limits climbed

The deal sits inside a broader compute build-out: Amazon (up to 5GW), Google/Broadcom (5GW), Microsoft/NVIDIA ($30B in Azure capacity), Fluidstack ($50B in U.S. infrastructure).

monday.com rebuilt its platform around agents that draft campaigns, qualify leads, triage tickets, and onboard employees.

Atlassian opened its Teamwork Graph (150 billion connections across people, work, and code) to any third-party AI agent that speaks Model Context Protocol.

Five major workplace platforms repositioned around agents in the same week: monday.com, Atlassian, Adobe, Twilio, and Google.

The timing was likely tied to Atlassian's Team '26 conference. But coordination at this scale makes the move structural — "AI features added" became "agents that execute work" across an entire category of software.

monday.com, used by 250,000+ organizations, rebuilt its platform around AI agents that take action, not just track it.

Native agents now draft campaigns, qualify leads, triage support tickets, process purchase requests, and onboard employees.

The decision in front of you is which workflows agents are allowed to act in, and which they aren't.

Gartner surveyed 350 executives at $1B+ companies.

80% of organizations using autonomous AI cut headcount. The cuts did not correlate with better AI returns.

75% of organizations that paused entry-level hiring in 2026 will pay a 15%+ premium for early-career professionals by 2030.

Three independent advisors converged on the same finding this week.

The most common AI business case template — "we'll pay for this with the headcount we don't need anymore" — doesn't generate the returns leaders are promising boards.

Gartner says the labor savings don't show up. Gartner separately says the talent debt does, with a 15%+ premium for early-career professionals by 2030 in organizations that paused entry-level hiring this year.

While the major AI labs are investing in the process of integrating AI tools into organizations from within, independent sources say that purely investing in AI won’t allow you to cut your workforce.

The market is continuing to pump incredible amounts of cash into the AI ecosystem, while the ones developing the core tech are signaling that in order to push adoption they’ll have to get directly involved to steer the ship.

It begs the question: What is your AI strategy and how are you measuring its effectiveness?

If you don’t have a strategy, start that process now. Don’t wait to have an opinion. Keep in mind the fact that you likely won’t cut headcount with AI adoption right now. So focus on leveraging it to increase efficiency and grow revenue with the team you have in place. AI implementation done right can unlock opportunities that were previously out of reach. Otherwise, you’re wasting that AI budget and falling behind on hiring at the same time.

Don’t cut your team of 50 to a team of 25 expecting to make up the gap with AI.

Make your team of 50 feel like a team of 500