The Anthropic IPO valuation is being built on a forecast that looks unusually far into the future. According to two people familiar with the company’s financials, Anthropic projects revenue of roughly $190 billion to $200 billion for 2028 — a figure not previously reported, and the anchor Wall Street is using to price what could become one of the largest public offerings on record.

The number sits against a very different present. As recently as May, the company publicised a revenue run rate of about $47 billion, reflecting its then-current pace of business. The gap between those two figures is the scale of growth investors are being asked to accept.

Why Bankers Are Reaching Into 2028

Valuing high-growth software companies on revenue multiples rather than profits is standard practice when a business has not yet settled into a mature earnings profile.

Stretching that forecast two years out is less usual. The people familiar with the process said it reflects both the speed at which the business is expanding and the difficulty of setting meaningful benchmarks for a company still spending heavily to build AI infrastructure.

That approach is not unprecedented in the current market. Analysts cited 2028 expectations ahead of another firm’s listing this year, and projections for SpaceX ran as far as 2029 before it went public at a record valuation in June.

The Search for Comparables

Ahead of Anthropic’s analyst day, three public companies are being weighed as reference points: Cloudflare, Palantir and SpaceX.

Comparables matter in IPO pricing because they establish what the market currently pays for businesses with similar growth characteristics, and they inform which multiple gets applied to a company’s own forecasts.

Where those companies trade, according to LSEG data:

• Palantir: roughly 53 times this year’s expected revenue, among the most richly valued names on Wall Street
• SpaceX: 41.6 times expected 2026 revenue
• Cloudflare: also 41.6 times expected 2026 revenue

Each offers a different angle. Palantir has become the standard reference for rapid growth paired with AI exposure. Cloudflare represents a high-growth software and infrastructure business. SpaceX illustrates a company priced substantially on expectations of future scale rather than present financial performance.

Why Earnings Are Being Set Aside

Mature companies are normally assessed on earnings or EBITDA, which reveal the underlying economics of the business.

For Anthropic, current EBITDA is not viewed as representative of what the company might eventually achieve. Spending on GPUs and other computing capacity, model training, inference and hiring is running at extraordinary levels. Those costs support the expansion — and the bet is that they shrink as a proportion of revenue over time.

The Growth Curve So Far

The trajectory the company has disclosed helps explain the willingness to look ahead.

Revenue run rate stood at roughly $9 billion at the end of 2025. By May it had passed $47 billion. Anthropic projected at least $10.9 billion in revenue for the second quarter of 2026 — more than double the preceding quarter — putting it on course for a first quarterly operating profit of $559 million.

The company has stated that its revenue run rate grew more than tenfold annually in each of the three years through early 2026.

The Core Assumption

Everything about the valuation rests on one proposition: that current spending is constructing a business capable of generating far higher revenue at considerably better margins.

Two mechanisms are expected to deliver that. Training and inference may become more efficient as the underlying technology improves. And personnel and general operating costs may occupy a smaller share of revenue as the business scales.

If either assumption fails, the multiple applied to a 2028 forecast becomes very difficult to justify.

A Note of Caution

Not everyone is convinced the trajectory holds indefinitely.

David Merkel, a principal at Aleph Investments, acknowledged the company could reach a $2 trillion valuation but questioned whether it would remain there over time. He framed the underlying issue as whether AI genuinely delivers the productivity gains attributed to it — the kind of question, he said, that anyone considering pricing or buying such an offering has to confront.

That scepticism is not confined to one investor. Heavy AI capital spending has already driven pullbacks in several widely held technology stocks in recent months, including some of the firms being used as comparables here.

What Could Go Wrong

Several risks sit underneath the headline number:

• Competition, which could compress pricing across the sector regardless of demand
• Compute costs, which may not fall as quickly as models assume
• Customer concentration, if a large share of revenue depends on relatively few contracts
• Regulatory developments, which remain unsettled across major markets
• Sentiment, since valuations built on distant forecasts fall fastest when confidence turns

A revenue multiple applied two years forward compounds all of these. Any slippage in the growth curve affects both the forecast and the multiple simultaneously.

Anthropic did not immediately respond to a request for comment.

Reading the Situation Fairly

There are two coherent ways to view what is happening.

One holds that the market is applying reasonable methods to an unusual company — that when revenue multiplies tenfold annually for three consecutive years, conventional trailing metrics genuinely fail to capture the business, and forward pricing is the honest response.

The other holds that forecasting revenue two years out for a young company in a fast-moving sector is closer to estimation than measurement, and that stacking a high multiple on top of an estimate produces a valuation with limited margin for error.

Both positions can be argued from the same set of facts. Which one proves correct will not be settled by the offering itself, but by whether the 2028 figure turns out to be conservative, roughly right, or considerably optimistic.