Anthropic discloses $17 billion Q3 revenue and $895.4 billion in commitments

Anthropic generated roughly $17 billion in revenue during the third quarter, according to an amended Form S-1 filing (the registration document companies must submit to the SEC before going public) submitted on Monday. The same filing disclosed purchase commitments and lease agreements totaling about $895.4 billion, a figure that anchors the capital intensity question hanging over one of the most anticipated public offerings in the technology sector.

The amended filing gives investors their first comprehensive look at the second half of Anthropic's financial year. The company's initial prospectus, unveiled last week and widely described as a step toward a potentially historic IPO, included financials only through the first half of the year. Monday's revision extends the record through Sept. 30.

What do the headline numbers show?

The nine-month picture is one of extraordinary growth paired with extraordinary cost. Anthropic reported $33.24 billion in revenue for the nine months ended Sept. 30, up roughly 1,200% from the same period a year earlier, when the company lost $25.1 billion over the comparable stretch.

The bottom line remains deeply negative. Anthropic reported a net loss of $104 billion for the nine-month period, with a net loss of $23.29 billion in the third quarter alone. The filing, however, attributes most of that nine-month loss not to operating cash burn but to a change in the fair value of convertible notes (debt instruments that can convert into equity) and derivative liabilities. That distinction matters for investors parsing the difference between accounting losses and actual operating economics.

The growth rate is the number most likely to anchor analyst models. A 1,200% year-over-year revenue increase over nine months places Anthropic among the fastest-scaling software businesses ever to approach public markets, and it does so while the company is five years old.

Why is $895.4 billion in commitments the figure investors will scrutinize?

The amended S-1 states that Anthropic has entered into purchase commitments and lease agreements with an aggregate committed spend of about $895.4 billion, "substantially all of which is related to compute and infrastructure." That single sentence describes a liability structure larger than the market capitalization of most companies in the S&P 500, and it quantifies the bet that AI labs are making on long-horizon capacity buildouts.

The filing explains the mechanism directly. "As our compute and infrastructure capacity grows, the associated compute and infrastructure expenses have increased," Anthropic said in the filing. The numbers bear that out: compute and infrastructure expenses rose by $14.4 billion for the nine months ended Sept. 30, an increase of 329% compared with the same period last year.

That 329% expense growth against roughly 1,200% revenue growth is the ratio the filing implicitly invites investors to examine. Revenue is scaling faster than the compute bill on these figures, though the $895.4 billion in committed spend means the expense line has a long, contractually locked runway ahead of it.

Who is Anthropic, and why does the IPO matter?

Anthropic is a five-year-old company best known for developing Claude, a family of AI models, and it primarily sells its products and services to enterprise customers. The company is capital intensive by design, largely because of the complex infrastructure and hardware required to train and run large models. The S-1 makes that structural dependence explicit in its risk factors.

The competitive frame is equally explicit. CEO Dario Amodei co-founded Anthropic alongside a handful of researchers who defected from OpenAI, now described in the filing's context as its chief rival. OpenAI confidentially filed its own prospectus with the SEC in June, but its executives have signaled a 2027 target for its public debut. Anthropic's amended filing, arriving ahead of that timeline, positions the two companies on a collision course for public-market capital and for the narrative of which lab can convert AI demand into durable economics.

The stakes extend past the two rivals. A successful Anthropic offering would provide the public markets with their most detailed financial window yet into the economics of frontier AI development, where revenue growth of four figures coexists with nine-figure quarterly losses and commitments approaching a trillion dollars. The offering would also test whether public investors will underwrite compute infrastructure commitments of this scale before the industry's return profile is fully proven.

What risks does Anthropic itself flag?

Anthropic has worked to position itself as more safety-conscious than its competitors, and the S-1 reflects that posture in its length: the filing includes more than 90 pages of risk factors. The company highlights risks including:

• Potential political headwinds facing the AI industry.
• Its complex infrastructure needs, now quantified in the $895.4 billion commitment figure.
• The difficulties that can arise from balancing safety objectives with commercial ones.

That last risk factor carries particular weight for a company whose brand identity rests on caution. The filing effectively concedes that the pursuit of revenue growth and the pursuit of safety practices can pull in different directions, a tension the risk disclosures address at length across those 90-plus pages.

The scale of the risk section also functions as a signal. Companies preparing for an IPO routinely use risk factors to pre-empt litigation over disclosures, but 90 pages exceeds the norm even by technology standards, and it aligns with Anthropic's stated positioning as the more careful actor in a field it has long argued is under-regulated.

What comes next?

The amended filing moves Anthropic one step closer to a debut that could rank among the largest IPOs on record, with pricing terms and timing still to come. Investors will now weigh the 1,200% revenue growth and the $17 billion quarterly figure against the $895.4 billion in committed spend, the $104 billion accounting loss, and a compute expense line growing at 329% — and the market's verdict on that trade-off will set a precedent for how OpenAI, targeting 2027, and every AI lab behind it gets valued.