The documents that reached Bloomberg reporters this week describe a company growing faster than any of its public peers. Anthropic, the artificial-intelligence startup behind the Claude chatbot, told investors in preliminary figures that second-quarter revenue exceeded $11.5 billion, more than 14 times the $787 million it booked a year earlier and more than double the $4.73 billion it reported for the first quarter. The company also recorded a positive adjusted operating profit, a line item that much of the AI industry has yet to reach.
The numbers now anchor a sales pitch. Reuters reported on Friday that Anthropic is betting its IPO valuation on a forecast of $190 billion to $200 billion in revenue by 2028, and that the company has begun early outreach to potential investors. Chief Financial Officer Krishna Rao is leading the meetings, according to the report, and a listing could begin as soon as this fall.
The two-year projection puts nearly the entire valuation question on a single growth curve. Anthropic said in May that its annualized revenue had passed $47 billion, and its full-year 2025 results crossed $10 billion only a few months earlier. The distance between where the company stands today and where its IPO story says it will be in 2028 is wide, and that is the gap investors will be asked to price.
The growth itself is real by any ordinary measure. Anthropic’s second quarter nearly doubled the first, and the year-over-year multiple of 14 reflects a customer base that moved from early pilots to production workloads. Enterprise clients now route large volumes of coding, customer service, and document work through Claude’s application programming interface, and the company has pushed subscription tiers into the mainstream consumer market. Revenue this large, this fast, changes the tone of the IPO roadshow: the question is no longer whether the product sells, but how long the slope can hold.
Costs are the counterweight. Training frontier models consumes clusters of graphics processors that cost billions to assemble, and serving millions of users requires data centers with power contracts that run for decades. Anthropic’s adjusted operating profit is a company-defined measure, and analysts caution that unadjusted results still show heavy investment. The company has answered with capital: Amazon and Google rank among its largest backers, and both provide cloud capacity that keeps compute bills inside the corporate family.
The IPO calendar adds its own pressure. A fall listing would place Anthropic in a window shared by other technology candidates, and bankers will test demand against the 2028 revenue forecast at the center of the pitch. People familiar with the process say the company is preparing for a range of outcomes, including a valuation that disappoints the private market’s expectations. Private investors have marked Anthropic’s shares up sharply in recent rounds, and a public price below those marks would sting the funds that led them.
Skeptics point to the crowded field. OpenAI, Google, and Meta all sell model access at aggressive prices, and open-weight releases have compressed the cost of inference across the industry. Anthropic’s answer has been specialization: Claude’s long-context and agentic tools have carved a niche in regulated industries such as law, finance, and health care, where accuracy claims matter more than raw price. Whether that niche supports a $200 billion revenue year is the question that the forecast begs.
Supporters argue the number is less extreme than it sounds. A company that nearly doubled quarter over quarter in the first half could plausibly compound from a $47 billion annualized base, and the AI market’s total spend is growing fast enough that even a modest share would cover the target. The forecast also lets Anthropic tell a story of rising operating margins: if revenue hits the projected range while compute costs rise more slowly, profitability would expand dramatically by 2028.
Regulators add a variable that no spreadsheet captures. Antitrust scrutiny of the cloud providers that fund AI startups has grown in both Washington and Brussels, and any forced change to Anthropic’s Amazon or Google relationships would alter its cost base. The company has also faced questions about safety obligations, data handling, and the export of frontier models, each of which carries the potential to slow a growth curve that the IPO valuation depends on.
The coming weeks will show how investors weigh the story against the numbers. Anthropic’s own disclosures, the tone of early meetings, and the reception of comparable listings will all feed the pricing process. What is clear is that the company has chosen to make its growth the entire argument, binding the valuation to a 2028 revenue figure that will be revisited quarterly for the rest of the decade. If the curve holds, the fall listing becomes the trade of the year. If it bends, the same forecast becomes the burden every new shareholder inherits.
For now, the company controls the narrative and the calendar. Rao’s meetings with potential investors have reportedly drawn full rooms, and the preliminary second-quarter figures give him a fresh set of numbers to show. The $11.5 billion quarter, the positive adjusted profit, and the 2028 forecast form one continuous claim: that Anthropic has become a scale business with a durable moat. The public market will soon deliver its verdict on whether that claim is worth the price its backers have already put on it.
