1. Stripe Acquires AI Gateway OpenRouter for More Than $7 Billion
Stripe has agreed to buy OpenRouter, the AI-model routing startup, for over $7 billion — a 5.4× markup on the $1.3 billion valuation OpenRouter set in its Series B just three months ago in May 2026. OpenRouter gives roughly 8 million developers a single access point to more than 400 models from OpenAI, Anthropic, Google, Meta and DeepSeek, and reportedly routed about 1.5 quadrillion tokens over the past year. The deal, reported August 16, plants the payments giant squarely in the AI infrastructure layer, where model choice and metered billing are converging.
2. Google's Gemini App Crosses 1 Billion Monthly Users
Sundar Pichai announced on August 11 that the Gemini app has passed 1 billion monthly active users, making it the fastest-growing product in Google's history and its 14th to reach the billion-user mark. The climb was steep: 400 million users at I/O in May 2025, 650 million by October, 900 million by May 2026, and now the full billion. That puts Gemini roughly level with OpenAI's ChatGPT, which crossed the same threshold in June 2026 — the consumer AI race is now a two-horse sprint at planetary scale.
3. Alibaba's Qwen3.8-Max: 2.4 Trillion Parameters, 1M-Token Context, Open Weights
Alibaba released Qwen3.8-Max in early August — a 2.4-trillion-parameter Mixture-of-Experts model (95B active) with a 1-million-token context window and native text, image and video input. It targets repository-scale coding agents, long-document knowledge bases and long-video indexing, and is priced aggressively at $2 per million input tokens and $6 per million output. With open weights following the API launch, it is the strongest open-weight challenger yet to the closed frontier labs.
4. Cloudflare Launches Kitesurf, a Browser Built for AI Agents — Not People
Cloudflare shipped Kitesurf, an agent-first web browser written in Rust that runs entirely in V8 isolates on Cloudflare Workers. It strips out everything a human needs but an agent doesn't — tabs, extensions, pixel-perfect rendering — and uses 3–7× less CPU and memory than Chromium for common agentic tasks like screenshots and HTML extraction. It speaks the Chrome DevTools Protocol, so existing Puppeteer and Playwright scripts work unchanged, and it's free while in beta. A sign the web is quietly growing a second, machine-native audience.
5. OpenAI Pushes GPT-5.6 Updates as Free-Tier Default After Steep Price Cuts
OpenAI published its GPT-5.6 August updates, and the model line is now the default for free ChatGPT users following an aggressive late-July price cut of roughly 80% on input tokens. The move squeezes the middle of the market: with Anthropic's Claude Sonnet 5 promo-priced at $2/$10 per million tokens through August 31 and Qwen3.8-Max at $2/$6, frontier-adjacent intelligence is in a full-blown price war heading into autumn.
6. AI Startups Raised $407 Billion in H1 2026 — Fewer, Bigger Checks
AI startups raised $407 billion in the first half of 2026, more than in all of 2025 combined — but deal count collapsed from 8,290 in 2025 to about 3,500 in H1, as investors concentrate capital into fewer, larger bets. August's big rounds fit the pattern: inference platform Fireworks AI landed $1.5 billion, Together AI raised $800 million, and enterprise-agent startup HappyRobot hit unicorn status with a $150 million Series C. Most capital is flowing to US labs, compute infrastructure, healthcare and regulated business tools.
7. DARPA and US Air Force Fly an F-16 Under AI Control
DARPA and the US Air Force confirmed that an AI agent has flown a modified F-16 fighter jet in live flight tests under the VENOM program, with a safety pilot in the cockpit able to retake control with the flip of a switch. The VENOM jets began airworthiness flights in June and moved to AI-controlled sorties in July, automating flight controls and sensors without changing the aircraft's core software. The fleet will now be used to test multiple AI agents in live-flight scenarios — groundwork for human pilots overseeing teams of autonomous, uncrewed aircraft.
8. INTERPOL: AI Now Involved in 55% of Cybercrime Across Africa
INTERPOL's African Cyberthreat Assessment 2026, drawing on data from 36 member countries, finds AI is enabling up to 55% of reported cybercrime on the continent, with losses more than doubling since 2024 to $484 million. AI is automating every stage of attacks — reconnaissance, phishing, extortion, evasion — while deepfakes and synthetic identities increasingly defeat biometric security. Scam centres now operate in 72% of surveyed countries, and the report warns that law-enforcement AI readiness remains alarmingly low.
// KEY TAKEAWAYS
The money is moving to the plumbing: Stripe paying $7B+ for OpenRouter, Fireworks and Together raising mega-rounds, and $407B poured into AI in six months — concentrated in fewer, bigger bets on infrastructure rather than apps. Meanwhile intelligence itself is commoditizing fast, with GPT-5.6, Claude Sonnet 5 and open-weight Qwen3.8-Max locked in a price war even as consumer AI hits planetary scale with both Gemini and ChatGPT at a billion users. And the stakes keep rising on both edges of the dual-use spectrum — AI now flies F-16s for DARPA and powers over half of Africa's cybercrime.