Anthropic Flips the AI Revenue Race Ahead of Dual IPOs

Anthropic reported preliminary second-quarter revenue above $11.5 billion, more than 14 times the year-earlier figure, and has told investors it projects $190 billion to $200 billion in 2028 revenue as it prepares a potential mega-IPO. OpenAI, still the consumer face of generative AI, has lifted its own annualized run rate above $40 billion. The two labs have turned their long rivalry into a straight revenue and listing race.

The split is no longer about model demos. It is about who converts usage into durable enterprise dollars faster, and which story Wall Street will pay the richer multiple for this fall.

The Numbers That Flipped the Race

By May 2026 Anthropic’s official run-rate had already passed OpenAI’s then-reported levels. The company stated that its run-rate revenue crossed $47 billion earlier that month after a $65 billion Series H that valued it at $965 billion post-money. OpenAI’s March funding round had set an $852 billion valuation on a lower run-rate near $25 billion.

Bloomberg later reported OpenAI’s annualized revenue had climbed above $40 billion by mid-August, roughly doubling the end-2025 pace, helped by coding tools, subscriptions and early advertising. Anthropic’s Q2 print still looks steeper: more than $11.5 billion versus $787 million a year earlier and $4.73 billion in Q1, with positive adjusted operating income for the first time.

Trackers have pushed Anthropic’s implied summer ARR higher still, into the high $60 billions or beyond, though the last company-confirmed figure remains the May $47 billion. The direction is clear: Anthropic compounded faster after both firms cleared the $1 billion ARR mark.

Claude’s Enterprise Cash Machine

Roughly 80 percent of Anthropic revenue now traces to enterprise and API use rather than pure consumer freemium conversion. More than 1,000 customers each spend over $1 million a year, a base that doubled in under two months after the February Series G. Eight of the Fortune 10 and about 70 percent of the Fortune 100 appear as clients in various tallies.

• Claude Code alone has been cited at multi-billion run-rates, becoming a core driver for technical teams.
• Products such as Cowork extend the same models into broader knowledge work.
• Claude sits on AWS, Google Cloud and Azure, with AWS remaining the primary training and cloud partner.
• Compute deals now include multi-gigawatt commitments with Amazon, Google/Broadcom TPUs and SpaceX GPU capacity.

That concentration produces higher revenue per user and stickier contracts. OpenAI still converts a massive free base into paid ChatGPT seats and is pushing enterprise harder, yet its historical mix stayed more consumer-heavy. One corporate-payments signal showed one in five Ramp businesses paying for Anthropic, up sharply year over year.

Bankers Price 2028 Not Today

Anthropic confidentially submitted a draft S-1 in early June. Sources told Reuters the firm is projecting roughly $190 billion to $200 billion in 2028 revenue. Bankers are applying enterprise-value-to-revenue multiples to those outer-year numbers rather than trailing or near-term figures, an approach used for other hyper-growth names such as SpaceX and Cerebras.

Comparables floated ahead of analyst day include Cloudflare and Palantir (both trading near 40-50 times expected revenue) and SpaceX. The stretch is obvious. Current margins remain pressured by GPU training, inference and hiring. Investors are underwriting the claim that revenue will outrun those costs as scale arrives.

Could they (Anthropic) get a $2 trillion valuation, yeah they could and I just wonder if it would stay there over time.

David Merkel of Aleph Investments voiced the open question many buyers share: whether the productivity gains justify multi-hundred-times earnings multiples once the stock is public.

Fall timing talk clusters around September-October for Anthropic, with OpenAI also aiming at a late-2026 window though some signals point later. Both want to list while AI enthusiasm remains elevated.

OpenAI’s Consumer Counterpunch

OpenAI still owns the broader brand and weekly active users measured in the hundreds of millions. Its monthly revenue pace hit roughly $2 billion earlier in the year and has accelerated further. Coding software, subscription tiers and nascent ads are adding new lines. Enterprise is no longer an afterthought; it now contributes a meaningful and rising share.

The company raised $122 billion at the $852 billion valuation in March with anchors including Amazon, Nvidia and SoftBank. It has filed its own confidential paperwork. Losses remain large, projected in the tens of billions for 2026 in some estimates, but the consumer flywheel keeps feeding data, distribution and pricing power.

Where Anthropic sells depth to a smaller set of high-value accounts, OpenAI sells breadth. That difference will show up in the S-1 risk factors and in how each company talks about path to sustained profitability.

Winners and Losers in the Dual IPO Window

Enterprise buyers already vote with budgets. The labs that lock multi-year, multi-million contracts capture switching costs and fine-tuning lock-in. Hyperscalers win either way: Amazon deepens its Anthropic primary relationship, while Microsoft, Google and others keep serving both models.

• Enterprise IT and procurement teams gain leverage as two near-trillion private companies compete on price, features and safety claims; earlier mid-tier price cuts by both labs already showed how quickly Chinese open-weight pressure forces adjustments.
• AI cloud and neocloud providers ride the same demand wave; AI cloud providers turning demand into prepaid growth illustrate how capacity sellers monetize the arms race.
• Public-market investors face two different bets: Anthropic’s denser enterprise base versus OpenAI’s scale and brand optionality. Early aftermarket volatility is likely if either misses the growth implied by 2028 forecasts.
• Employees and secondary sellers at both firms get liquidity events that could set new compensation benchmarks across AI.

Chip and memory suppliers sit upstream of both. The same capacity scramble that produced multi-gigawatt deals also explains why substrate and HBM shortages keep appearing in the broader supply-chain coverage.

How the Safety Split Became a Revenue Split

Anthropic was founded by former OpenAI researchers who left over safety and governance priorities. For years that positioning looked like a commercial handicap. In 2026 it reads more like a sales asset. Large regulated buyers appear willing to pay a premium for the constitutional-AI framing, auditability and enterprise controls Claude emphasizes.

1. Late 2024-early 2025: both labs cross $1 billion ARR; Anthropic still far smaller.
2. End 2025: Anthropic run-rate near $9 billion.
3. February 2026: Series G lifts Anthropic to $14 billion run-rate and $380 billion valuation.
4. April 2026: Anthropic crosses ~$30 billion run-rate, overtaking OpenAI’s then-cited figures on some trackers.
5. May 2026: Series H at $965 billion; official $47 billion run-rate; Q2 path to first profitable quarter.
6. June 2026: confidential S-1 filed; OpenAI follows with its own paperwork.
7. August 2026: Anthropic Q2 >$11.5 billion and 2028 forecasts surface; OpenAI run-rate tops $40 billion.

The safety narrative never disappeared. It simply stopped being the only story. Enterprise trust and coding-agent traction turned the original cultural difference into a revenue advantage. OpenAI’s broader distribution remains a formidable counterweight, especially if advertising and consumer ARPU keep rising.

The Public Market Test Ahead

Both companies still burn heavily on compute. Both face the same open questions around model commoditization, regulation and whether agentic products stay sticky once every lab ships comparable coding and workflow tools. The Nvidia security alliance that left the frontier labs out is one reminder that infrastructure players are building their own coalitions.

Anthropic enters the listing conversation with higher recent private valuation, faster historical compounding after the $1 billion mark, a denser enterprise mix and an explicit multi-year revenue target that forces investors to underwrite 2028. OpenAI enters with unmatched consumer mindshare, a larger user base and its own accelerating run-rate now above $40 billion.

The race is no longer theoretical. The first of the two to price will set the multiple that the second must beat or justify. Enterprise customers will keep writing the checks that decide which story ages better once the private-market optimism meets public-market scrutiny.