GitHub Copilot changed how it charges you. Every AI tool is watching.

Hi AI Futurists,

GitHub Copilot switched to usage-based billing on June 1, and developers are not happy. The frustration is understandable. But Copilot is not the story here. It is the preview. Every flat-rate AI subscription you pay for is quietly running the same calculation right now. Let's take a look.

Our agenda:

Best,
Lex Sokolin

P.S. Hit reply with suggestions or feedback!

2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.

Download the Toolkit

Our friends at the The AI-Driven Enterprise (AIDE) Institute, are an independent research organization that measures, benchmarks, and accelerates the transition to AI-first organizations worldwide.

They share some incredible research on the state of AI, and recently published a 2026 S&P 500 report with their findings.

We asked them to put together some exclusive takeaways for Future Blueprint readers around the growing AI adoption divide.

Here’s what they had to say:

Despite the hype, corporate America’s AI transformation is still in its early innings. A new analysis of all 500 companies in the S&P 500 by the AIDE Institute finds nearly four in ten (38.8%) companies are “Emerging Adopters,” trailing their industry peers on both AI leadership and hands-on execution. The signals beneath the surface are just as telling: 44.2% of these corporate giants hold no AI patents, and three-quarters have leadership teams where AI fluency is still the exception. Even in tech, 40% of companies lag. Among the best-resourced companies on earth, AI adoption is a divide — not a done deal.

Explore the full rankings

📋 Anthropic files confidential IPO prospectus with SEC (CNBC). Closed $65B on May 28. Filed four days later, targeting October at a $965B valuation.

💰 Apollo and Blackstone arrange $36B chip debt deal for Anthropic (crypto.news). The largest chip-financing debt deal in history, used to buy Google TPUs through an SPV. Private credit just became AI infrastructure.

🔧 Microsoft launches MAI-Code-1-Flash and MAI-Thinking-1 at Build (CNBC). Two new coding and reasoning models, cheaper than OpenAI, benchmarked against Anthropic. Microsoft's AI chief says he is now "less concerned" about Google, Meta, and OpenAI.

💻 GitHub Copilot token billing goes live with no model fallback (AI News). Developers are calling it "a joke." Your monthly AI coding bill is now unpredictable.

🤖 First confirmed live LLM-agent cyberattack: AWS database exfiltrated in under two minutes (The Hacker News). An LLM agent pivoted through four steps from an exploit to a database with no human in the loop. Proof of concept is now just concept.

🏗️ SoftBank commits €75B to build 5 GW of AI data centers in France (TechCrunch). Masayoshi Son flew to Paris. EDF's nuclear grid is the pitch.

🚀 Cognition's Devin raises $1B at $26B valuation, hits $492M ARR (The Agent Report). 50% monthly growth, Mercedes, NASA, and Goldman as customers. Nine months ago Devin was a $10B company.

🔒 Project Glasswing expands Anthropic's Mythos to 150+ organizations in power, water, and healthcare (Investing.com). The AI that finds hidden software flaws now lives inside critical infrastructure.

📱 OpenAI retires GPT-4.5 and o3, upgrades GPT-5.5 Instant (gHacks). GPT-4.5 launched four months ago and is already gone. GPT-5 superseded it on every benchmark that matters.

🏢 OpenAI launches DeployCo, a standalone enterprise deployment company (OpenAI). The model lab is becoming a systems integrator. Enterprises do not buy models, they buy implementation.

GitHub Copilot is used by over 15 million developers and, until June 1, it cost a flat monthly fee: $10 for individuals, $19 per user for businesses. Simple. Predictable. That changed this week. GitHub switched to token-based usage billing, which means you now pay based on how much you actually use, and the rate depends on which underlying AI model your request hits. The monthly price did not change, but it now buys a credit allowance rather than unlimited access. When the credits run out, you either pay more or stop. The model fallback that used to quietly route heavy users to a cheaper option is gone. Developers on GitHub's community forums called it "a joke," with most complaints focused not on the amount but on the unpredictability.

Flat subscriptions were always a subsidy. When Copilot was $10/month, GitHub was absorbing whatever the actual compute cost of your usage turned out to be. That model made AI tools feel like software you could budget for once and forget. Token billing ends that. It shifts the cost variance back to the user and makes AI feel more like a utility: the more you use, the more you pay, and the bill comes at the end of the month. GitHub Copilot is the first major consumer-facing AI product to make this shift at scale, but it will not be the last. OpenAI has charged API users this way for years. The question was always when it would hit the products that millions of people use daily, not whether it would.

For founders with AI in their products, and for anyone who pays flat monthly fees for AI tools right now, this week is a preview of a coming conversation. Every AI tool running on frontier models has a cost structure that does not scale cleanly with flat pricing. The tools still holding at flat rates are either absorbing the cost, running lighter models under the hood, or waiting until adoption is high enough that users will accept the change rather than leave. When usage goes up, and with agents running more tasks automatically it will, flat fees break. Copilot moved first because Microsoft was willing to absorb the political cost of changing. Others will move when their margins force it. If you have AI subscriptions where "unlimited" is in the fine print, it is worth reading those terms again.

Build Club Campus — Free AI learning community for professionals who want to get genuinely good at building with AI

Empromptu AI — Build full-stack AI-native apps in minutes without writing code

TimeTuna — Scheduling pages with custom video backgrounds for client-facing bookings

This 158-page research report provides the first comprehensive taxonomy of public companies, private ventures, and tokenized protocols building the infrastructure for autonomous AI systems. Compiled by Lex Sokolin, former Chief Economist at ConsenSys, fintech strategist at Autonomous Research, and current Managing Partner at Generative Ventures, this report delivers institutional-grade analysis of 100+ companies across 14 critical infrastructure layers. Learn more here.

That’s all for today, folks!