Why is Google cutting off Meta?

BIG TECH

Even Google doesn't have enough Google AI to go around.

Google has started limiting how much of its Gemini models Meta can use, Reuters reported over the weekend. Meta had been leaning heavily on Gemini through Google Cloud to power internal workloads like content moderation and safety automation, and Google decided it was time to pull some of those tokens back.

The reason isn't strategic so much as it is mechanical. Google doesn't have the compute to keep selling Gemini at the rate Meta wants to buy it. Sundar Pichai said as much on the Q1 2026 earnings call, admitting that Google Cloud revenue would have been higher if the company could actually meet demand. The numbers back him up:

+ Google Cloud backlog hit more than $460 billion, nearly doubling in a single quarter. Just over half converts to revenue in the next 24 months — meaning Google has years of committed workloads it physically can't deliver yet.

+ Capex for 2026 was raised to $180 to $190 billion, with Pichai signaling 2027 will be "significantly" higher. Even at that scale, demand is still pulling ahead.

+ Gemini's API is now processing more than 16 billion tokens per minute, up from 10 billion the prior quarter. Meta is one of the largest of those customers.

The logic from Google's side is straightforward: it would rather sell those tokens to enterprise customers paying for Gemini Enterprise than keep subsidizing a competitor's product roadmap.

And Meta is very much a competitor. After the rough reception to Llama 4, Meta launched Muse Spark as its new flagship but has been quietly using Gemini to keep things like moderation and safety pipelines running at scale. When your supplier is also your rival and your supplier runs short, you're the first one rationed.

The deeper issue is that the bottleneck isn't really chips anymore — it's power. Transformer availability and local utility capacity are what's actually gating how fast anyone can bring new compute online. Google can buy all the TPUs it wants. It still has to get them powered up.

If Google is rationing one of the biggest AI labs in the world, nobody else is safe from the same call. The story of 2026 was supposed to be which model is smartest. The story it's actually becoming is which company can plug in enough of them to matter. Compute scarcity is starting to do what model benchmarks couldn't, which is decide who gets to build at scale and who has to wait in line. Meta just learned what that line feels like.

PRODUCTS

The most conservative white-collar industries are turning out to be the most aggressive early adopters of coding agents. Which is not how anyone called it.

Inside OpenAI, the lawyers, finance team, and recruiters now use Codex, the company's coding agent, as their primary AI tool. According to OpenAI's own breakdown, those teams crossed over around April and now generate more than 85% of their output tokens through the tool. The median lawyer at the company is running 12x more work through Codex than they did in November.

And the pattern shows up well beyond OpenAI:

+ Walleye Capital: The hedge fund says 100% of its roughly 400 employees use Claude Code, Anthropic's equivalent of Codex.

+ Citadel: Analysts are using Claude inside Excel to build coverage models and separate signal from noise.

+ Big Law: Firms are running internal "vibe coding" programs where non-engineers spin up custom tools for case prep.

These are industries that usually touch workflow changes last — heavily regulated, billing by the hour, and not historically rewarding engineers in suits. Watching them pull coding agents deeper into the daily routine than some engineering teams is the genuinely surprising part.

The key insight is that "coding agent" turned out to be a misnomer. People aren't shipping production software. They're building one-off scripts, automating spreadsheet work, and standing up internal tools that used to require begging an engineering team for time. About a quarter of non-engineer Codex work is coding-shaped. The rest is research, drafting, and analysis.

Not everyone is sold. Chris Bedi, ServiceNow's Chief Customer Officer, has been openly skeptical of measuring AI progress by token volume. "It's almost like measuring a restaurant based on how many ingredients they buy," he said. "You don't measure a restaurant that way." And there's a quality concern backing that up: Stanford's legal-tech researchers found that teams using agents without domain-specific verification introduced 12–18% more compliance revisions than traditional methods. The lawyers vibing their way through case prep are also generating cleanup somebody else has to catch.

The expected story about AI in the enterprise was that engineers would adopt first and everyone else would follow years later. What's happening instead is that the suits are running the same tools as the engineers, just for entirely different reasons. Finance and legal didn't fall for Codex because they want to ship software. They fell for it because the agent lets one person do work that used to require asking three other departments. If that pattern holds, the industries most exposed next are the ones built around middle layers nobody wants to call anymore.

WORKFORCE

Companies spending the most on AI aren't the ones cutting jobs. They're the ones hiring the most.

A new joint analysis by Revelio Labs and Ramp covering more than 21,000 US firms found that companies in the top third of per-employee AI spending grew headcount 10% in the two years after adopting — with entry-level roles growing 12%. The bar to qualify as a "heavy adopter" wasn't even high: about $30 per employee per month, roughly the cost of a ChatGPT Team seat.

Ramp economist Ara Kharazian's takeaway is worth bookmarking: "If you are reading headlines where CEOs blame layoffs on AI, be skeptical."

He has reason to say that, because the headlines are very real. According to the May 2026 Challenger report, AI was cited as the reason for 38,579 layoffs in May alone — 40% of all announced cuts. Through May, AI has been blamed for 87,714 job cuts in 2026, already blowing past the 54,836 attributed to AI in all of 2025.

So both things are true at the same time: companies investing in AI are growing, and companies citing AI as the reason for layoffs are also growing. The interesting question is what's actually driving the cuts. A closer look at the Challenger data suggests many of these companies are restructuring through mergers and bankruptcies — cuts attributed to M&A alone are up more than sixfold year over year. AI just makes a better headline than "we overhired in 2022."

There's a generational wrinkle, too. The Revelio finding of 12% entry-level hiring growth only captures firms actively investing, not the ones retrenching — and other research suggests early-career roles are taking the hardest hit elsewhere. Stanford's Erik Brynjolfsson, who has been tracking AI's labor impact for years, put it bluntly: "We are flying blind into one of the most consequential periods in world history."

The cleanest read on all of this is that AI is doing what every major technology shift has done before, which is reward the companies leaning in and expose the ones using it as a cost-cutting excuse. The firms hiring more are the ones using AI to expand what they can do. The firms blaming AI for layoffs are mostly firms that needed to cut anyway and found a quieter way to say it. If you want to know which side a given company is actually on, look at what they're spending, not what they're announcing.

IN OTHER NEWS

+ OpenAI and Anthropic limit new AI models to Trump-approved customers during cybersecurity review

+ GitHub Copilot's first token-billing invoices arrive and developers report costs jumping 10x to 50x

+ Taiwan raids Super Micro offices in expanded Nvidia AI chip smuggling probe

+ Meta secretly flooded rival chatbots with 45,000 crisis prompts from fake teen accounts

+ Etched raises $800M and locks in $1B in sales for its transformer-only AI chip

+ Humanoid robot maker Apptronik raises $1B at $5B valuation , Mercedes-Benz invests

+ UK MP sues Elon Musk's xAI over AI-generated sexualized images of her

+ EU Council gives final approval to simplified AI rules , bans AI-generated sexual deepfakes

WHO'S HIRING IN AI

+ Anthropic — Field Marketing Manager

+ OpenAI — Program Manager, Outreach & Audiences (Asia Pacific)

+ Northrop Grumman — AI Systems Engineer (Principal Level)

+ Airbnb — Principal AI/ML Researcher, Bayesian & Reinforcement Learning

AI OR REAL?

Which image is real?

Option A | Option B

AI TOOLS

+ Gemini: Google's AI assistant now creates personalized images by pulling from your Photos, Gmail, and Search history — just say "design my dream house" and it already knows your taste

+ Perplexity: The AI search engine launched Computer for Counsel, a legal-specific agent that routes tasks across 20+ AI models and links every answer to the actual ruling or statute so lawyers can verify citations in one click

+ Claude Code: People are now using Anthropic's coding tool to edit full videos — Fable 5's own launch video was edited entirely inside the terminal, no video editor needed

+ Retell AI: The voice AI platform launched Conductor, a review tool that lets businesses see exactly where their AI phone agents fail and automatically fix conversation flows

+ Waves ILLUGEN 2.0: The audio plugin maker released a text-to-sound engine that generates studio-quality loops and samples ready for professional music production