Apple Opens Siri to Third-Party AI in Biggest Platform Shift Since the App Store
Apple's WWDC 2026 keynote delivered what may be the company's most consequential AI announcement yet: starting with iOS 27, users will be able to replace Siri's underlying intelligence with third-party models including Anthropic's Claude, Google's Gemini, and OpenAI's ChatGPT.
The new Extensions framework transforms Apple's walled garden into a competitive marketplace for AI assistants. Rather than relying on a single bilateral deal — as it did with ChatGPT integration in iOS 18 — Apple is opening a standardized integration layer that any qualifying AI provider can join through the App Store. Users can configure their preferred AI in the Apple Intelligence & Siri settings, and a dedicated Extensions section in the App Store will serve as the discovery hub.
Craig Federighi set the tone during the keynote: "We believe privacy in AI is non-negotiable." The new Siri, powered by Google Gemini as the default, is described as more conversational, context-aware, and capable of visual intelligence — and now has its own dedicated app alongside its traditional system-wide integration.
iOS 27 also brings significant quality-of-life upgrades: Photos load 70% faster, AirDrop transfers are 80% quicker, and system-wide AI dictation now handles spelling, punctuation, and capitalization corrections. Perhaps most intriguingly, developers have already spotted references to "foldState" and "angleDegrees" in the beta code — fueling speculation about a foldable iPhone reveal at Apple's September event.
The keynote also marked Tim Cook's farewell, as he confirmed his transition to John Ternus as CEO on September 1, 2026.
OpenAI and Anthropic Both File for IPOs, Setting Up the AI Market's Defining Moment
The two most prominent AI labs are heading to public markets nearly simultaneously. Anthropic filed its draft S-1 with the SEC on June 1, followed by OpenAI's confidential filing on June 8 — setting up what could become the most closely watched pair of tech IPOs since the dot-com era.
The numbers tell a striking story. Anthropic has reached a $1 trillion valuation on secondary markets, with 123% year-to-date appreciation and near-profitability in its first quarterly earnings. OpenAI, meanwhile, was last valued at $852 billion but has seen just 11.3% appreciation this year and recently missed revenue and user targets.
OpenAI projects it will burn through $85 billion in 2028 despite doubling sales, with $122 billion in annual computing costs, and doesn't expect to be cash-flow positive for at least four more years. The company emphasized flexibility in its filing, noting "there are things we want to do that are likely easier as a private company."
With 900 million weekly active users, OpenAI's scale is unmatched — but Anthropic's filing first could constrain OpenAI's pricing power with investors. The race to Wall Street mirrors their race in the lab.
Microsoft Launches Seven In-House AI Models, Challenging Its Own Partners
Microsoft unveiled a family of seven new MAI models built entirely in-house — a bold move that puts the company in direct competition with the very AI labs whose models it hosts on Azure.
The flagship MAI-Thinking-1 is a medium-sized reasoning model that matches leading competitors on software engineering benchmarks and was preferred over Anthropic's Sonnet 4.6 in blind human evaluations. Notably, Microsoft emphasized it was built from scratch without distillation from third-party models — a pointed differentiation from competitors accused of training on others' outputs.
MAI-Code-1-Flash, a 5-billion-parameter agentic coding model, is deeply embedded in GitHub Copilot and VS Code, offering Haiku-class performance at lower cost. The lineup also includes MAI-Image-2.5 for text-to-image generation, MAI-Transcribe-1.5 with state-of-the-art accuracy across 43 languages, and MAI-Voice-2 for multilingual speech synthesis.
The models are available through Microsoft's Foundry platform as well as OpenRouter, Fireworks, and Baseten — signaling Microsoft's ambition to build a full-stack AI offering independent of its OpenAI partnership.
Washington Drafts First Comprehensive AI Law as White House Orders Model Reviews
The US is entering a new chapter of AI governance. On June 4, Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a discussion draft of the Great American Artificial Intelligence Act (GAAIA) — the first bipartisan attempt at comprehensive federal AI regulation.
The bill spans four titles covering frontier AI governance, workforce protections, cybersecurity, and international cooperation. It would require mandatory audits for major AI companies and establish protections for workers displaced by AI. Critically, it would also preempt state AI regulations for three years on model development — though deployment rules remain at states' discretion.
This follows President Trump's June 2 executive order requiring frontier AI companies to submit cutting-edge models for a 30-day government review — a significant shift from the administration's earlier hands-off posture. OpenAI CEO Sam Altman responded with a detailed regulatory blueprint, while Trump floated the unprecedented idea of the government taking equity stakes in major AI companies.
Experts note the proposals focus heavily on national security and catastrophic risks while leaving everyday concerns largely unaddressed — job displacement, youth safety, and data privacy remain on the margins. Colorado's comprehensive AI law takes effect June 30, adding further urgency to the federal-state jurisdictional battle.
Google Pays SpaceX $920M Monthly for Emergency GPU Access
In a deal that underscores the extraordinary demand for AI compute, Google has agreed to pay SpaceX $920 million per month from October 2026 through June 2029 for access to approximately 110,000 NVIDIA GPUs — a contract worth roughly $11 billion total.
Google described the arrangement as "a short-term, timely agreement to ensure bridge capacity" driven by unexpected demand for its Gemini Enterprise agent platform. The fact that the world's largest AI compute owner needs to rent additional GPU capacity externally speaks volumes about the current state of the AI infrastructure race.
SpaceX has quickly emerged as a major compute infrastructure player — it previously secured a $1.25 billion monthly deal with Anthropic for similar services. The two companies are also reportedly in talks about building orbital data centers, a concept that would have seemed like science fiction just a few years ago.