Loading today’s stories
Product1 distinct publisher2 min readPublished
The figure Anthropic is expected to show IPO investors describes a pool 12.5 times the S&P 1500 tech sector's revenue. Its own 2028 forecast reaches about 0.67% of it.
The Product Desk · Product desk
Compiled by The Product DeskSomething wrong?How this is made
Annualise what Anthropic actually collected and the gap becomes measurable. The $11.6bn booked in the second quarter [14] runs to roughly $46bn a year, about 0.15% of a $30tn pool [2]. The company's own projection for 2028, $190bn to $200bn, reaches about 0.67% at the top end [15][3]. The growth is real: revenue more than doubled in the quarter and is up more than fourteenfold in a year [14]. What the addressable market adds is the other 99.3% [3], and nothing in the disclosure record covers it.
The reason the total can be that large is a change of denominator. A software market is capped by what companies spend on software; this one is built from the full scope of work the models could complete, which counts human labour [6]. Payroll is a far bigger pool than IT budgets, so the ceiling moves by an order of magnitude without anyone having to forecast a price. Read it as a statement of intent rather than a projection. A vendor whose addressable market is labour will eventually quote against the cost of the person doing the task, and the only checkable version of that claim is task by task, with a named party confirming the work was finished.
There is recent evidence on what the total by itself buys. SpaceX filed what it called the "largest actionable" market in "human history" at $28.5tn in May [3], of which $26.5tn came from AI rather than rockets [9], roughly 93% of the figure [4]. The shares jumped, gave it back, fell below $105 in intraday trading in early August, and now trade around the $135 offering price [12]. Three months on, the investor who bought the biggest addressable market ever printed has made nothing [13].
Aswath Damodaran, who teaches finance at New York University, told the Journal before that offering that the AI portion was "reaching the end of what's plausible and pushing beyond" [10]. Anthropic's number would sit $1.5tn above it [11]. It would also be five times the $6tn Uber counted in 2019 by valuing the mileage of every personal car and public transport service on earth [7][5], a figure Wall Street called aggressive at the time [7]. WeWork cited $3tn in an offering that never happened [8].
None of this is official. Anthropic has published nothing, said nothing publicly about a $30tn market, and did not immediately respond when Reuters asked; the people describing the figure said the numbers could still change [18]. The auditable number in the story remains $11.6bn [14].
Ranked by verification strength, evidence, and original report placement.
Anthropic is likely to tell IPO investors that its potential revenue opportunity runs above $30tn, reported by Corrie Driebusch for The Wall Street Journal citing people familiar with the matter.
A total addressable market describes the annual revenue a company could theoretically capture at 100% market share, and bankers build it from industry data and their own models. The company is not claiming it will collect the money, but the pool is the argument for the price.
SpaceX put $28.5tn in front of investors before its record listing, calling it in its May filing the "largest actionable" market in "human history".
The 191 technology companies in the S&P 1500 booked $2.4tn in revenue last year, according to FactSet data cited by The Wall Street Journal.
Anthropic arrives at the figure by looking at the full scope of work AI models could complete, a framing that counts human labour rather than software spending, according to the same people who spoke to the Journal.
When Uber went public in 2019 it claimed a $6tn market opportunity, reached by counting the mileage value of every personal car and public transport service on earth; Wall Street called that aggressive at the time.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Concrete comparators, second-hand core claim
The comparison figures are specific and checkable (FactSet revenue for 191 S&P 1500 tech companies, SpaceX's May filing language and $26.5tn AI share, Damodaran's on-the-record quote, Anthropic's Q2 revenue and 2028 forecast range). The central claim, however, is an unpublished number attributed to anonymous people familiar with the matter and relayed by a single aggregating publisher, with Anthropic silent and the figure explicitly subject to change.
Real revenue base, negligible against the pool
Uptake evidence exists and is strong in absolute growth terms — $11.6bn in a quarter, more than fourteenfold year on year — but it covers roughly 0.15% of the described market, and even the top of the 2028 forecast reaches about 0.67%. No customer, deployment, or labour-substitution evidence in the cluster supports uptake at the scale the TAM implies.
Market framing far ahead of collected revenue
The reported pool is roughly 12.5 times the entire listed US tech sector's revenue and about 150 times Anthropic's annualised run rate, while the nearest precedent — SpaceX's 93%-AI $28.5tn TAM, called past the edge of plausibility by Damodaran — has produced no shareholder return in three months. The article itself is the corrective rather than the amplifier, and Anthropic's actual revenue trajectory is genuinely steep, so the gap is large but not total.
Number built to price an offering
The figure is a pre-IPO pitch artefact: bankers construct TAMs from their own models, the pool is the argument for the price, and per the reporting it also justifies the valuation and the infrastructure spending that valuation must support, guides product priorities, and frames rivalry with OpenAI and Google. Its unpublished, deniable status ahead of a filing due within weeks sharpens the incentive to test a maximal number in private.
Sound arithmetic, thin sourcing
The internal arithmetic and comparators are consistent and can be recomputed from figures given, and the article is candid that nothing is official. Confidence is capped by a single-publisher cluster relaying anonymous sourcing about an unpublished figure that the sources say may change, with no primary filing available.
build
Grok 4.6 lands in Copilot two days after launch, and the model picker becomes a procurement problem1 distinct publisher
product
Baidu's AI line grew 25 percent and still lost the arithmetic1 distinct publisher
security
The nationalization argument is really a vendor-continuity memo1 distinct publisher
invest
Google Ships Flash Instead of Pro While OpenAI Loses Its Two Best Operators1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 25, 2026