Anthropic's $1.5 Trillion IPO: What It Means for How We Work
Three years ago, most of the AI tools in our studio were novelties. Today they sit inside how we write, code, audit brands and build reports. So when the company behind Claude goes public, it's worth paying attention — not as a stock tip, but for what it tells us about the tools we all rely on.
Anthropic priced its shares at US$83, above the indicated range of US$70–78, raising about US$81.3 billion and valuing the business at roughly US$1.5 trillion. Add a separate US$26 billion private placement taken up by companies including Nvidia, Google, Microsoft, AMD, Broadcom, Micron and TSMC, and it's the largest stock-market debut on record.
The real value for the rest of us is the paperwork. Anthropic's IPO prospectus gives the clearest look yet at what frontier AI actually costs to build and run. This is a balanced summary of what's in it — and none of it is financial advice.
A company built around a safety promise
Anthropic was founded in 2021 by Dario and Daniela Amodei and a group of former OpenAI researchers. From the start, the pitch was that safety research and capable models should be developed together, not traded off.
That thinking shows up in how the company is structured:
- A public benefit corporation. Under Delaware law, the board must balance shareholder returns with a stated purpose: the responsible development and maintenance of advanced AI for the long-term benefit of humanity.
- The Long-Term Benefit Trust. An independent group of financially disinterested trustees holds a special class of shares and elects a majority of the board.
- Limited say for new shareholders. Founders hold shares carrying about 50.1% of voting power, so people buying Class A shares get economic exposure with very little influence.
Supporters say this protects safety decisions from quarterly pressure. Critics say it asks investors to fund enormous spending with no real way to rein it in. Both views are reasonable, and the market will test them over time. We covered the practical side of this balance in Using AI Effectively and Ethically.
Fast growth, heavy costs
The numbers are big in both directions:
- Revenue. Up roughly twelve-fold in 2025 to almost US$4.6 billion, with US$17 billion reported for the third quarter of 2026 alone.
- How it earns. About US$3.8 billion of 2025 revenue came from usage-based billing, with subscriptions at US$789 million. The company expects usage to stay the substantial majority.
- Losses. Operating losses more than doubled in 2025 to over US$8 billion.

Training and running these models takes huge amounts of computing power, mostly through Amazon and Google's cloud platforms. The money raised gives Anthropic room to keep building, but it also raises the bar: from now on, every quarterly report will be read for signs the costs are coming under control.
Partners that are also rivals
One of the more telling disclosures is how much depends on two companies. In 2025, 47% of sales went through Amazon and Google's cloud marketplaces. Both are major investors, both supply computing power, and both build competing AI models.
For businesses using these tools, the lesson is familiar: don't lock everything into one supplier. It's the same thinking we bring to any tech stack, and it's covered in How AI is Changing the Game for Australian Business.
The case for, and the case against
Putting the optimism and the caution side by side:
- For: Claude is widely used for coding, reasoning and long documents; enterprise customers tend to stay; and the raise funds years of development.
- Against: Model prices keep falling as rivals catch up; revenue is concentrated through two partners; losses are large; and public shareholders have little control.
- Unknown: How regulators in the US, Europe and Australia will treat AI, cloud bundling and model risk over the next few years.
The next 12 months

Here's what we'll be watching:
- From chat to coworker. Expect more AI that works through longer tasks in the background — coding, research and operations — rather than answering one prompt at a time.
- Margin scrutiny. Investors will look for cheaper, faster models and efficiencies that narrow the losses.
- Regulation. Competition and safety regulators are already looking closely at AI partnerships and disclosures.
- Business integration. The gains will go to businesses that connect AI to their own data, products and customers.
That last point is the one that matters most locally, and it's the thinking behind Join the Dots: Web, Brand, Ads & AI SEO. If you're choosing tools now, pick them by the job, not the feature list.
Our take
Anthropic's listing marks the point where frontier AI stops being a research project and becomes infrastructure, priced and questioned like any other utility. For Australian businesses, that's mostly good news: the tools will keep improving, and there'll be far more transparency about the companies behind them.
Use the capability, keep an eye on the costs, and build on your own data and craft — not just someone else's software.
Sources
- Anthropic, PBC — Class A Common Stock Prospectus (November 2026)
- Reuters — Anthropic valued at $1.5 trillion as IPO raises record $81 billion
- Reuters — Anthropic IPO prospectus lays bare deep dependence on Big Tech partners
- TechCrunch — Anthropic sets IPO price range of $70 to $78 per share
Want AI working properly in your business?
We help Australian businesses put AI to work across design, marketing and reporting — sensibly, and on your own data.
