AI: Anthropic warms up Investors for 2026 mega-AI IPO. AI-RTZ #1176

The mega-AI IPO parade has long had a marching order this AI Tech Wave. Elon’s SpaceX/xAI went first earlier this year, blew past two trillion dollars in early trading, and then reminded everyone what the trip from private to public feels like when the shares sold off.

Anthropic is next, targeting September or early October. And OpenAI now likely follows as late as next year. We have been tracking this tight IPO clock for months. And Anthropic is getting ready with preliminary meetings with institutional investors.

The WSJ lays it out in “Anthropic Tries to Shore Up Investor Confidence Ahead of Blockbuster IPO”:

• “Anthropic is meeting with potential investors to shore up confidence ahead of a planned IPO that could launch in September or early October.”

• “Investors have pressed Anthropic on challenges including cheaper Chinese AI systems, Trump administration tensions and data-center opposition.”

• “Anthropic told some investors it plans to push further into healthcare and biology”.

The core near-term mission is straightforward. The company, currently valued in the private markets at $965 billion, is quietly meeting potential investors ahead of what could be the largest IPO of all time, taking the hard questions now, and developing the story that will work best when it goes out officially this Fall.

Per the Journal, investors are pressing back on three primary worries:

1. The recent popularity of cheaper open source AI systems made in China,

2. Tensions with the Trump administration,

3. And the growing backlash against data center construction across the US.

Topics we’ve covered in detail here on AI-RTZ and daily podcast ARD.

Anthropic’s answers: it is hyperfocused on cutting-edge models, most users want the most intelligent AI available at any given time, and Chinese systems generally trail the frontier by months. The company also told some investors it plans to push further into healthcare and biology AI uses, work that could soften some of the negative public sentiment around AI.

The numbers underneath are why investors lean in anyway. Anthropic said in May that its run-rate revenue had topped $47 billion, powered by the breakout success of Claude Code. The company has forged a flurry of new compute deals with SpaceX and Google in recent weeks (at premium to market rates). Usage keeps running hot enough that the constraint has been compute, not customers, with intermittent outages this year to show for it.

My take: of the three mega-AI IPOs, Anthropic remains the most relatively attractive. Not the cheapest, not the safest, but the one whose story requires the fewest leaps of AI faith. That view rests on three things, and it is worth being explicit about them since the bear case is real too (Not stock advice).

The first is singular focus. Anthropic is an enterprise and AI Coding company, full stop. Claude Code took off in a way that we flagged early in its ‘Vibe Coding’ moment, and it is the reason the WSJ can call the company the front-runner in the AI race. There are no consumer hardware side quests, no video feeds, no social apps, no AI device ambitions. One business, sold to businesses, billed by the token a la carte by the buckets.

The second is the text-first bet. Ben Thompson’s Stratechery piece this week draws the contrast sharply. Google’s DeepMind under Demis Hassabis pursued world models and multimodality, the grand vision of AI that sees and simulates. Anthropic went the other way: Claude outputs text only, by design, and the payoff shows up in exactly the place enterprises are paying for today, AI Coding and long-running agentic workflows. Thompson goes further: with Hassabis moved upstairs, Google itself may now be pivoting toward Anthropic’s more text- and code-centered approach. When your biggest rival starts steering toward your lane, you are in the right lane.

The third is the compute cost curve. Anthropic built on Google’s TPUs and Amazon’s Trainium years ago, and has not been dependent on Nvidia’s CUDA for years. Per the analysts Thompson cites, more than a fifth of TPU shipments through late 2027 are being sold directly to Anthropic, which is now buying chips outright for its own data centers, converting marginal compute costs into capital costs. In a world where intelligence gets priced like a commodity, the low-cost producer wins the margin argument, and Anthropic has quietly positioned itself as one.

Now the other side of the ledger. Anthropic has had the tougher road with the US government of any frontier lab. We wrote about ‘The Blip 2.0’, the three-week freeze on its frontier models, when Washington’s scrutiny landed on Anthropic first and hardest. The company’s safety-first posture reads as friction in this administration, even as ‘slow and steady’ becomes the new US AI speed for everyone. The irony is thick: the lab that volunteered for guardrails got the least credit for them.

But that same safety focus is also the shield. Against open source models and the China AI Tigers shipping cheaper systems weekly, Anthropic’s pitch to enterprises and governments is trust. Models that are tested, aligned, and accountable, from a company that treats safety as a product feature rather than a press release. That is a moat commodity models cannot easily cross, and it is the story that plays best with the institutional investors now in these meetings.

And the valuation math hangs over all three IPOs. Trillions of dollars in AI compute spend, as we covered in our supply and demand series this week, rests on the assumption that demand keeps rising precipitously. SpaceX’s post-pop selloff is the fresh reminder that public markets mark that assumption to market daily. The price Anthropic sets, and how it trades after, will set the stage for how investors value every top AI developer.

The part I find most instructive is the rehearsal itself. These preliminary discussions are not fundraising, they are story development.

Hear the objections early, tune the narrative, then go out officially with answers already in the deck. It is what we saw when Anthropic made three filed IPO take-offs back in June, and what the valuation debates since have demanded: a story that survives contact with skeptics.

OpenAI, meanwhile, threw in the ‘Kitchen Sink’ on costs ahead of its own offering, and is now expected as late as next year, refocusing on enterprise and AI Coding and trimming the side quests. The Anthropic tortoise laps the OpenAI hare once more: the narrow, patient story gets to price first, and the sprawling one gets to explain itself second.

The mega-AI IPOs started filing while the market irons were hot.

This Fall we find out what the market thinks they are worth when the irons meet the water.

An impact of the AI Tech Wave worth tracking into this Fall. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)