Anthropic quietly files for a near-trillion-dollar IPO, the EU publishes its first deepfake labeling rulebook, and India's national AI infrastructure crosses 38,000 GPUs.
Anthropic files for IPO at a near-trillion-dollar valuation, higher than OpenAI
Anthropic has confidentially submitted its IPO filing while closing a financing round that put its valuation at $965 billion, surpassing OpenAI for the first time. The company is also in talks to lease US data centre space with backing from Google. A public listing in late 2026 would make it the largest AI IPO in history.
Why it matters: This is the clearest sign yet that AI labs are becoming publicly traded infrastructure companies, not just research organisations. The money that flows in after an IPO will accelerate everything (compute, hiring, product velocity) across the whole industry.
The EU just published its first rulebook for labeling deepfakes and AI-generated content
The European Commission released a Code of Practice setting out how AI providers must mark and disclose AI-generated content, text, images, audio, and video, including deepfakes. It's voluntary now but is designed to be the compliance template before the EU AI Act's transparency rules come into full force. Companies already operating in Europe are expected to sign on.
Why it matters: This is the EU formalising what "responsible AI content" looks like in practice. Voluntary today does not mean optional tomorrow, companies that ignore this framework will find it harder to argue good faith when enforcement begins.
India's AI Mission hits 38,000 GPUs and launches a ₹1,000 crore semiconductor push
India's IndiaAI Mission has now provisioned over 38,000 GPUs for startups, with another 20,000 planned soon. The Union Budget 2026–27 announced India Semiconductor Mission 2.0, dedicating ₹1,000 crore to chip research, training, and workforce development. National skilling programmes (SOAR, YUVAi, and AI fellowships) are expanding alongside the infrastructure build-out.
Why it matters: India is building AI infrastructure at national scale and running workforce programmes in parallel. The window to build a product, career, or research project on this subsidised compute is open right now, but it won't stay open indefinitely.
OpenAI launches a $150M partner network to push AI into global businesses
OpenAI officially launched its Partner Network, committing $150 million to help companies around the world accelerate enterprise AI adoption. The programme covers deployment support, transformation frameworks, and access to OpenAI's latest models, aiming to turn enterprise buyers into long-term platform customers.
Why it matters: OpenAI is shifting from selling API access to selling a full transformation path. If your organisation is evaluating AI rollout tools, this partner route is now a structured option, and a signal that enterprise AI is moving from pilot phase to mainstream deployment.
AI now consumes as much electricity as an entire country, and regulators are taking notice
The International Energy Agency warns that AI-related electricity consumption could hit 1,000 TWh by 2026, roughly equal to Japan's total annual usage. Data centre power demand may double between 2022 and 2026. At the same time, the World Economic Forum notes AI is being deployed to accelerate renewable energy solutions and cut industrial emissions, framing it as both the problem and a key part of the solution.
Why it matters: AI's energy footprint is no longer an abstract concern, it's showing up in electricity prices, planning decisions, and climate negotiations. Countries and companies making major AI investments will also have to account for the grid capacity required to run them.
Labcorp launches an AI app that explains your blood test results in plain English
Labcorp released MyLabcorp, a HIPAA-compliant mobile app that uses OpenAI's reasoning models to let patients ask questions about their lab results through a conversational chat interface. Instead of a confusing printout, you get a plain-language explanation and the ability to follow up with questions.
Why it matters: This is practical AI for everyday people, not developers. Health literacy is one of the most underserved use cases for AI, and a regulated, compliant tool doing it at scale is exactly what the technology was supposed to be for.
Colorado's AI law takes effect in 14 days, and the fight to stop it isn't over
America's first comprehensive state AI law becomes enforceable on June 30, covering automated systems that affect hiring, housing, healthcare, and credit. Federal negotiations are still ongoing about whether states can pass such laws at all, with some proposals to preempt state rules entirely in exchange for weaker federal ones. What happens in the next two weeks sets the pattern for AI regulation across every US state.
Khanmigo reaches 18M students + Avataar builds culturally-aware video AI for India
Khan Academy's AI tutor Khanmigo now serves 18 million students globally and is widely considered the gold standard in classroom AI tools. Separately, Indian startup Avataar has secured new funding for its video AI platform built specifically for India, designed to be cheaper, faster, and culturally aware, supporting local languages and regional contexts at scale.