Quick word before we get into it. I know the name says ChatGPT, but honestly, this has never really been about one chatbot. ChatGPT was the spark. What I am watching is the whole AI era, every company, every tool and every change to the way we work. And this fortnight proved the point nicely: the biggest moves did not come from OpenAI at all. They came from Anthropic, Microsoft and Google.
This was the fortnight AI grew up and went looking for a stock ticker. Anthropic raised a staggering $65 billion, quietly passed OpenAI to become the most valuable AI startup on the planet, and filed to go public. SpaceX lined up $30 billion of Google compute to help fund its own listing. The frontier is being built for Wall Street now, and compute contracts have become the new collateral.
The models did not sit still either. Anthropic’s Opus 4.8 took back the top spot, Microsoft rolled out seven of its own models to cut its reliance on everyone else, and Google handed its sharpest reasoning mode to power users. Even the robots had their moment, with Computex calling them the single biggest opportunity in tech. Bold claim, hard to dismiss.
Then there is the stuff you will actually feel. ChatGPT is turning itself into a do-everything superapp, folding agents, coding and image generation into one place, and its memory is even learning to update itself as your life changes. The way we actually use these tools day to day is shifting faster than the headline deals suggest.
But the story that stuck with me most was a quieter one. Nearly half of companies now admit their AI rollout has been a let-down, even while the model makers raise record sums. That gap, between the polished demo and what AI actually does in everyday work, is the real headline of 2026. Tellingly, this was also the fortnight Pope Leo XIV used his first encyclical to say AI must be disarmed, and 272 experts put the odds of a catastrophic outcome at one in ten. Worth sitting with.
Closer to home, Australia is living all of this at once: a $155 billion data-centre boom and a new $10 billion build in South Australia, set against Sydney fretting about its water and Microsoft pointing out that the real gains might be on a mine site, not a laptop.
My take? The winners this year will not be whoever has the biggest model or the biggest round. They will be the teams who quietly make AI useful in the unglamorous corners of real work, and the leaders honest enough to admit where it is not landing yet. The money has moved to Wall Street. The value still has to be earned on the ground.
Anthropic closed a $65 billion Series H at a $965 billion post-money valuation on 28 May, the largest raise in its history and very likely its last as a private company. The round vaults it past OpenAI’s $852 billion valuation to make the maker of Claude the world’s most valuable AI startup for the first time.
The round was co-led by Altimeter, Dragoneer, Greenoaks and Sequoia, with Capital Group, Coatue, GIC and Blackstone joining and chipmakers Micron, Samsung and SK hynix taking strategic stakes. Run-rate revenue has crossed $47 billion, and the Wall Street Journal reports a 130% surge that should deliver Anthropic’s first operating profit. Within days, the company confidentially filed its IPO prospectus with the SEC.
It announced all of this on the same day it shipped Claude Opus 4.8, its new top-scoring model, and signalled it would more widely release systems on par with its cautious cyber model Mythos. The capital funds an enormous compute build-out spanning Amazon, next-generation Google TPUs and SpaceX’s Colossus sites.
For two years OpenAI led on valuation and consumer mindshare. Anthropic has now flipped the ranking. What makes the moment notable is how it happened: the lead came from enterprise adoption of Claude Code and Cowork, not a viral consumer hit. That is a slower, stickier base, the kind that shows up as run-rate revenue rather than download charts, and it is exactly the story public-market investors want to hear.
The raise also resets the IPO race. With a confidential S-1 now filed, Anthropic, OpenAI and SpaceX could all reach the public markets within months of each other, turning 2026 into the year private AI valuations finally get tested by real shareholders. Bundling the funding, a new flagship model and an IPO filing into a single day was no accident; it is a message that capability, revenue and demand are all compounding at once.
The bottom line: watch the valuation Anthropic targets on its roadshow, whether that first operating profit actually lands this year, and how regulators treat a $965 billion company that openly holds back its most capable models on safety grounds.
This fortnight the centre of gravity shifted to the capital markets. Anthropic’s $65 billion raise and IPO filing, SpaceX’s $1.75 trillion listing underwritten by roughly $75 billion of contracted Google and Anthropic compute, and DeepSeek’s first external round all point the same way: the frontier is being financed for public shareholders, and multi-year compute contracts have become the collateral. Microsoft building its own models, rather than only renting them, is the same instinct in a different form.
Yet the most revealing number was not a valuation. It was the 48% of companies now calling their AI rollout a disappointment. The capital has moved to Wall Street. The value still has to be earned in everyday work, and that is the gap worth watching for the rest of 2026.