Stripe Acquires OpenRouter for Over $7 Billion
In the biggest AI infrastructure deal of the year, Stripe has finalized its acquisition of OpenRouter for more than $7 billion, Bloomberg reported on Saturday. The deal represents a staggering 5.4x markup from OpenRouter's $1.3 billion Series B valuation just three months ago.
OpenRouter, founded in 2023, operates as an AI model gateway — a routing layer that helps developers switch between hundreds of AI models from providers like OpenAI, Anthropic, Google, and Meta. CEO Alex Atallah has described the company as "the Stripe for AI," and now it's literally becoming part of Stripe.
The acquisition signals Stripe's ambition to become the payments and billing backbone for the AI economy. As companies increasingly consume AI through APIs — paying per token, per request, per minute of compute — Stripe sees model routing as a natural extension of its payments infrastructure. By owning the layer that decides which model handles a request, Stripe can also meter, bill, and settle the transaction.
OpenRouter's investors, including Sequoia, Andreessen Horowitz, Menlo Ventures, and Alphabet's Capital G, stand to see significant returns on what was already one of the fastest-growing AI startups of 2026.
Anthropic Posts First Profit, IPO on the Horizon
Anthropic's growth trajectory continues to defy expectations. The Claude maker reported preliminary Q2 2026 revenue of over $11.5 billion — a 14-fold increase from the $787 million it generated in Q2 2025. Even more significant: the company posted positive adjusted operating income for the first time, hitting profitability two years ahead of its own internal schedule.
The numbers are remarkable even in context. Anthropic's revenue jumped from $4.73 billion in Q1 to over $11.5 billion in Q2, suggesting that demand for Claude models is accelerating rather than plateauing. The company's annualized revenue run rate now exceeds $47 billion.
With a confidential S-1 filed on June 1, Anthropic is positioning for what could be a fall 2026 IPO. Morgan Stanley, Goldman Sachs, and JPMorgan Chase are preparing the offering, with the valuation expected to hinge on projected 2028 revenue targets of $190–200 billion. If it proceeds, Anthropic would become one of the first major private AI companies to go public.
Google DeepMind Undergoes Major Leadership Shake-Up
Jeff Dean, Google's chief scientist and one of the most influential figures in modern AI, is leaving the company after 27 years. Dean, along with fellow Google legends Sanjay Ghemawat, Oriol Vinyals, and Quoc Le, is co-founding Discovery Loop, a startup aiming to automate the experimental loops of scientific research using AI.
The departure triggered a broader reorganization at Google DeepMind. Demis Hassabis is stepping back from day-to-day leadership to become chair of DeepMind and Alphabet's chief scientist, while Koray Kavukcuoglu takes over as SVP of Google DeepMind. Alphabet shares fell about 4% on the news.
Google is a founding investor in Discovery Loop and will serve as the startup's cloud partner, softening the blow somewhat. But losing Dean — who co-authored the foundational MapReduce and TensorFlow papers, among many other contributions — represents a symbolic turning point. The talent drain comes at a challenging moment for Google, which has also seen Nobel Prize-winning researcher John Jumper leave for Anthropic earlier this summer.
Meta Open-Sources Muse Glimmer: A 30B Model That Runs Locally
Meta continues to push its open-weight AI strategy with the release of Muse Glimmer, a 30-billion-parameter multimodal model licensed under Apache 2.0. Released on August 10, Muse Glimmer is specifically optimized for local agentic workflows — always-on AI agents that can run on consumer hardware.
The engineering achievement is notable: Meta's team compressed the model's footprint from the expected ~55 GB to under 20 GB, making it runnable on a Mac or PC with a single consumer GPU. Despite its compact size, Muse Glimmer ranks #1 among local models on the MCP Atlas benchmark (75.5) and scores 51.2 on SWE-Bench Pro.
The release accompanies Muse Code, Meta's terminal-based coding agent powered by Muse Spark 1.2, and a 6,500-word essay from Mark Zuckerberg titled "The Future is for Everyone" making the case for open AI development. Meta also announced it will open the weights of Muse Spark 1.2, its most advanced model.
EU AI Act Transparency Rules Now in Full Force
As of August 2, the European Commission began enforcing the transparency provisions of the EU AI Act — the world's most comprehensive AI regulation. Article 50 obligations now require:
- Chatbot disclosure: AI systems interacting directly with people must clearly identify themselves as non-human
- Deepfake labeling: AI-generated or manipulated audio, images, and video must be labeled
- Content marking: AI-generated text on matters of public interest must carry machine-readable watermarks
- Emotion recognition disclosure: Systems performing emotion recognition or biometric categorization must inform users
The rules apply immediately to all in-scope systems regardless of when they were launched, though providers of generative AI systems already on the market have until December 2, 2026, to comply with the marking and detection requirements. Noncompliance can trigger fines of up to €15 million or 3% of worldwide annual turnover.
The enforcement comes as China also moves to regulate AI agents specifically, with its Implementation Opinions on AI Agents — widely described as the first national policy to treat AI agents as a distinct regulated category — taking effect on July 15.