It has been an unusually heavy week in AI, and the common thread is money: a near-trillion-dollar IPO filing, an €87bn data-centre pledge, a landmark chip launch, and two more lawsuits landing on AI companies. Below are the eight stories that drove the most discussion between 28 May and 5 June 2026, with a short summary of each and why it matters. Figures and filings move quickly in this market, so dates are noted throughout — re-check anything before you rely on it.
1. Anthropic files confidentially for a near-trillion-dollar IPO
On 1 June, Anthropic confidentially submitted draft IPO paperwork to the SEC, potentially beating rival OpenAI to a Wall Street debut as soon as this autumn. The Claude maker recently closed a funding round at a $965bn valuation — topping OpenAI's $852bn from late March — and said its revenue run-rate has reached about $47bn, up from $10bn a year earlier. The honest caveat, raised by Fortune: a confidential filing is still only an option to go public. There are no audited public numbers yet and the timing depends on SEC review and market conditions. Even so, it sets up the first real price-discovery moment for a frontier AI lab, which is why the whole sector is watching.
2. Microsoft Build: an always-on agent called Scout, and seven in-house models
At its Build conference on 2–3 June, Microsoft unveiled Scout, an "always-on" personal work agent that handles meeting prep, scheduling and routine tasks across Teams and Outlook without being asked. It also released a family of seven in-house "MAI" models, led by MAI-Thinking-1, its first reasoning model trained from scratch. Scout is being pitched not as another chat box but as a persistent agent with its own identity, memory and permissions — something IT teams will need to manage like infrastructure rather than a feature. The strategic subtext drawing the most comment is that, by building its own frontier models, Microsoft is steadily reducing its dependence on OpenAI.
3. Nvidia's RTX Spark pushes AI onto the PC
At Computex in Taipei on 1 June, Jensen Huang unveiled the RTX Spark superchip, an Arm-based PC processor built with Microsoft that arrives this autumn in Windows machines from Dell, HP, ASUS, Lenovo and MSI. It is Nvidia's first serious move into the laptop processor market long ruled by Intel, AMD, Qualcomm and Apple, and the news sent shares of those rivals lower. Huang said Nvidia and Microsoft would "reinvent the PC". The wider story is AI moving from the data centre to the "edge" — running capable models directly on your device instead of the cloud, which has implications for cost, privacy and what offline AI can do.
4. SoftBank pledges up to €75bn for AI data centres in France
At the Choose France summit on 30 May, SoftBank committed up to €75bn (around $87bn) to develop 5GW of AI data-centre capacity in France — its largest AI infrastructure investment in Europe. The first phase is €45bn for 3.1GW in the Hauts-de-France region by 2031, across sites at Dunkirk, Bosquel and Bouchain, with state utility EDF providing a former power-plant site and Schneider Electric partnering on power infrastructure. It is a marker of how central cheap, low-carbon energy and industrial land have become to the AI race, and of Europe's push to host more of the compute its companies depend on rather than renting it from the US.
5. GitHub Copilot switches to token billing — and developers push back
On 1 June, GitHub moved every Copilot plan to usage-based billing, replacing premium request units with "AI Credits" charged on the tokens you consume at each model's API rates. TechCrunch called it the end of Copilot's "golden age" for smaller users, and developers shared projections of monthly bills jumping from $29 to $750, or $50 to $3,000, on heavy agentic workflows — though basic code completions stay free. The reaction is genuinely split: some developers welcome it as fairer, since light users keep predictable caps while heavy users pay for what they burn. Either way, it is the clearest sign yet that the era of "unlimited AI for a flat fee" is closing as the real cost of running models reaches customers.
6. CNN sues Perplexity over copyright
On 28 May, CNN sued Perplexity, accusing the AI search company of unlawfully copying and distributing its content — CNN's first lawsuit against an AI firm, and believed to be the first by any television network. The complaint, filed in New York, alleges Perplexity scraped more than 17,000 CNN stories, photos and videos, and says licensing talks last year failed. Perplexity's response was characteristically blunt: "You can't copyright facts." The suit adds to a growing pile; by late May, around nine organisations — including the New York Times, Dow Jones, the Chicago Tribune, Encyclopedia Britannica, Merriam-Webster and Reddit — had active cases against Perplexity. These are allegations, and the underlying fair-use questions remain unsettled in court.
7. Florida sues OpenAI and Sam Altman over child-safety claims
Florida became the first US state to sue OpenAI and chief executive Sam Altman, filing on 1 June and alleging the company knew ChatGPT was unsafe — especially for minors — while marketing it as safe and reliable. The suit seeks to hold Altman personally liable and follows a criminal investigation tied to a 2025 mass shooting; the state's attorney general says penalties could reach "billions". These are allegations, not findings. OpenAI responded that ChatGPT is used by hundreds of millions of people for legitimate purposes and that it "works continuously to strengthen safeguards" against misuse. The case sharpens a question that will define the next phase of AI regulation: who is liable when a general-purpose AI is linked to real-world harm.
8. Suno raises $400m at a $5.4bn valuation
AI music generator Suno announced a $400m round on 3 June, valuing it at $5.4bn — more than double its valuation from roughly seven months earlier. The raise, led by Bond Capital, lands as the platform reports more than two million paying subscribers and over seven million songs generated daily. The tension running through it: Suno still faces copyright lawsuits from major labels and acknowledges training on copyrighted songs, which it argues is fair use. It has since settled with Warner Music and signed a licensing deal — its first with a major label. The investment is a clear bet that AI music is here to stay, even with the law around it unresolved.
Frequently asked questions
What was the biggest AI story this week?
Anthropic's confidential IPO filing on 1 June, at a $965bn valuation, was the most-discussed — it could become one of the largest technology listings ever and may put Anthropic on public markets ahead of OpenAI.
Did GitHub Copilot get more expensive?
For heavy users, yes. From 1 June, Copilot charges for the tokens you use rather than a flat monthly rate, so intensive agentic workflows can cost far more. Basic code completions remain free, and light users keep predictable caps.
Why is Perplexity being sued so often?
News publishers allege it copies and redistributes their journalism without permission or payment. CNN's 28 May suit is the latest of around nine active cases, reflecting a wider fight over whether AI search can use publishers' content under fair use.
The bottom line
The thread running through the week is that the money and the lawsuits are arriving together: record valuations and infrastructure pledges sit alongside an expanding wall of litigation over copyright and child safety. The most practical signal for everyday users is the GitHub Copilot change — the first clear sign that cheap, flat-rate AI is ending as the true cost of running these models lands on customers. If you budget for AI tools, plan for usage-based pricing to spread, and keep a record of what your heaviest workflows actually cost.