Invest1 distinct publisher2 min readUpdated
Ramp data from 70,000 companies puts Fable 5 under about 11% of spend on Anthropic tools, according to the Financial Times. Budget owners have found the cheapest lever: stay off the frontier.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
What Ramp can see is spend passing through the payment systems of 70,000 companies, which is not the same thing as Anthropic's revenue [2]. The report does not say whether large negotiated enterprise agreements sit inside that sample or outside it. So treat the roughly 11% as a read on the visible mid-market rather than a line from a prospectus.
Even read generously, though, the number is a ceiling rather than a snapshot: Fable 5 spend has yet to surpass about 11% at any point [2]. And the window it was measured across was not a clean one. The model's June debut was cut short when the White House forced Anthropic to withdraw it on national security grounds, and the Trump administration only later permitted a relaunch [6]. Analysts and investors told the FT that price and performance, not the interruption, are what decide which model customers pick [6].
That is the part budget owners should register. The substitution here happens inside the vendor relationship, not away from it. Nobody has to rip out a contract or requalify a supplier to cut an AI bill; they move work to an older tier that the FT's sources say handles the majority of business demand [3]. Downgrading is the lowest-friction cost lever in the stack, and roughly 89% or more of observed Anthropic spend is now sitting on the cheap side of it [1].
The awkwardness for the listing is that Miles Clements of Accel, which has put nearly $1 billion into Anthropic, is the one narrating this as normal. Most people do not need to operate at the frontier, he said, and the era when customers reached only for frontier models "was not a durable era" [4]. As a description of the market that is plausible. As a line from an investor weeks before an offering, it concedes that the newest and presumably highest-priced product is not where the money goes. Anthropic declined to comment to the FT and had not responded to PYMNTS [5].
One more thing worth straightening out. The report frames this as the largest IPO on record, at a valuation of at least $2 trillion, surpassing SpaceX, which raised $75 billion at the outset and $86.2 billion with the overallotment [7][8]. Those are different measures: a company valuation against money actually raised. The $2 trillion figure is about 23 times SpaceX's final proceeds [2], and how much of it converts into a raise is the number that has not been published.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
The Financial Times reported on Sunday, Aug. 23, that Anthropic's U.S. customers are using lower-cost alternatives to its most powerful AI model, raising questions about the company's aggressive spending ahead of its expected offering.
Spending on Anthropic's Fable 5 model has yet to surpass roughly 11% of overall expenditure on Anthropic's tools, according to the FT citing data from 70,000 companies collected by payments firm Ramp.
The FT said the pattern runs against corporate customers gravitating toward the most powerful models, and that analysts and Anthropic investors attribute the trend mainly to Fable's high price and older models handling the majority of business demands.
Miles Clements, a partner at Accel, which has invested nearly $1 billion in Anthropic, said "Most people don't need to operate at the frontier," and that the period when customers tended to opt only for frontier models "was not a durable era."
Anthropic declined to comment when reached by the FT, and had not replied to PYMNTS at the time of publication.
Fable 5's debut in June was interrupted when the White House forced Anthropic to withdraw the model over national security concerns; the Trump administration has since allowed a relaunch, but analysts and investors say price and performance remain the larger factor in model choice.
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.
One quantified datapoint, two removes from the source
The cluster contains a single aggregator article relaying a Financial Times report that itself cites Ramp payments data from 70,000 companies. The central number is specific and quantified, which lifts it above rumor, but there is no primary dataset, no methodology, no Anthropic comment (the company declined), and the IPO valuation and timing are hedged expectations rather than confirmed terms.
Broad tool usage, thin frontier-tier uptake
Adoption here is unusually well quantified for a single-source story: spend appears across a 70,000-company panel, so Anthropic tooling itself is widely bought, but the frontier model in question has never exceeded roughly 11% of that spend, implying about 89% or more sits on cheaper models. The score reflects measured but narrow uptake of the specific model at the center of the story, with no seat count, token volume, or named enterprise deployment disclosed.
Record-IPO framing outruns the demonstrated demand
The narrative pairs a hedged 'largest IPO on record, at least $2 trillion' expectation — roughly 23 times SpaceX's ultimate $86.2 billion raise, comparing a valuation to proceeds — with a single secondhand spend datapoint. Claims are overstated relative to what the evidence establishes: no revenue or margin data links frontier spend share to the valuation case, and neither the offering terms nor the dataset methodology is disclosed. The gap is moderate rather than extreme because the core 11% figure is specific and attributed.
Named investor and pre-listing framing shape the telling
The only on-record voice is a partner at Accel, an investor with nearly $1 billion in Anthropic, whose framing that 'most people don't need to operate at the frontier' recasts weak frontier uptake as a natural market maturation. Anthropic itself declined to comment, so no counterweight is present, and the report is published in the weeks before a potential record offering when spend-mix narratives are directly price-relevant. The payments-data provider also gains visibility from being the cited source.
Plausible and specific, but unverified and single-sourced
One aggregator relaying one outlet relaying one vendor dataset, with no company confirmation, no methodology, and no corroborating publisher. The spend-share claim is internally coherent and specific enough to be actionable directionally, and the regulatory episode is concrete, but the IPO valuation and timing remain speculative. Confidence sits below the midpoint accordingly.
invest
The first AI IPO writes the comp sheet, and Anthropic is holding the pen1 distinct publisher
invest
The AI trade's weak link is the buyer: Anthropic's best model took 6% of its tokens1 distinct publisher
product
Card data, not vendor decks: the enterprise AI split is unsettled, not settled1 distinct publisher
product
Lightspeed's Project Mercury puts the firm on both sides of the OpenAI price1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 23, 2026