Anthropic's revenue growth and IPO preparations

The headline number is the trajectory itself: a $9 billion annualized run rate at the end of 2025 climbed to $47 billion by May 2026 and then to $65 billion by the end of July - a more than sevenfold jump in about seven months ^[1]. The quarterly detail is even sharper. Preliminary Q2 2026 revenue came in above $11.5 billion, a 14-fold increase over the $787 million Anthropic generated in Q2 2025, and more than double the $4.73 billion it reported for Q1 2026 ^[2]. Layered on top of that acceleration, Anthropic disclosed its first positive adjusted operating income in Q2 2026, a profitability milestone that gives the company a materially different pitch to public-market investors than a lab still burning cash at scale ^[3]. Anthropic's own backers are now modeling a $100-120 billion run rate by the close of 2026 ^[1]- which would mean the company more than doubles again in the second half of the year, on top of the sevenfold gain already booked.

Anthropic's growth numbers are landing in the middle of a live contest for who controls the largest IPO in history. Morgan Stanley, Goldman Sachs and JPMorgan Chase have been tapped as lead underwriters, with Goldman and Morgan Stanley separately jockeying against each other - and against bankers working OpenAI's parallel IPO process - for the prestige 'lead left' bookrunner slot ^[4][5]. That competitive backdrop matters because OpenAI is chasing the same public-market moment: its own revenue run rate sits around $40 billion, well behind Anthropic's $65 billion, and Anthropic's earlier arrival at operating profitability gives it a cleaner story to tell first ^[3]. Yet the person closest to Anthropic's own numbers, CFO Krishna Rao, has reportedly avoided putting a number on the table at all - leading early investor meetings without discussing a specific valuation target, even as outside reporting settles on a $2 trillion figure that would eclipse SpaceX's $1.77 trillion valuation from its June 2026 listing ^[6][7].

Underneath the trillion-dollar framing sits a narrower, more technical argument. Some analysts suspect Anthropic may be booking large upfront enterprise cash payments for annual token allocations immediately as revenue, ahead of the compute costs those tokens will eventually consume - a timing question that bears directly on how durable the reported growth actually is ^[8]. That question feeds into the bigger one: the $2 trillion valuation case reportedly rests on a projected 2028 revenue of roughly $190-200 billion, a figure Anthropic has not issued as formal guidance ^[9]. Getting there from today's $65 billion run rate would require more than quadrupling revenue in about two years - an assumption Aleph Investments' David Merkel calls into question, noting he isn't sure a $2 trillion price tag 'would stay there over time' and pressing on whether AI's claimed productivity gains are real enough to justify it ^[9]. Not every voice on the sell side agrees the market is overpricing Anthropic, though: one investor cited via the Financial Times argues that at an 800%-per-year growth rate, even a conservative 30x revenue multiple would put fair value well above $2 trillion, suggesting the debate cuts both ways ^[7].

Commentary tracking this story outside the traditional financial press has skewed noticeably more skeptical than the underlying wire reporting. The loudest independent voices frame the growth and profitability figures as at least partly an artifact of accounting and timing choices rather than pure demand - arguing that some of the reported jump reflects Anthropic recently starting to charge enterprise customers rates that had previously been discounted or absorbed, and that a temporary compute-cost break contributed to the profitability milestone. That framing lands as sharply critical, with the eventual S-1 predicted to invite far more scrutiny than the investor update has so far received. Retail investor discussion echoes the same tension: debate has centered on whether Anthropic has a durable moat against cheaper, fast-improving open-weight Chinese models, and whether a still-private company projecting $190-200 billion in 2028 revenue is describing a realistic plan or marketing a story to bankers. Notably, the wire reporting that originated the $65 billion figure itself has stayed neutral and matter-of-fact, meaning the skepticism is concentrated in commentary and analysis rather than in the underlying disclosure - a gap worth watching as the actual S-1 becomes public.