$16.8B Revenue and an Adjusted Profit: With a Closed-Door Investor Day, Anthropic Moves One Step Closer to the 'Largest IPO Ever'
Anthropic’s IPO journey has taken another big step forward. On October 14, the Claude developer held a closed-door pre-IPO investor day at its San Francisco headquarters, disclosing its latest financials to prospective institutional investors. The numbers that leaked afterward immediately became the focus of tech and capital markets.
The core takeaways: Q3 revenue reached $16.8 billion, up about 46% from Q2’s $11.5 billion; even more critically, Anthropic told investors it achieved an adjusted profit in Q3, with positive cash flow for the quarter. That stands in stark contrast to the market’s established image of “Anthropic, the loss-making giant.” And all this anticipation points to one clear goal—pursuing what could be history’s “largest IPO ever,” with a target valuation exceeding $2 trillion.
Against this backdrop, I can’t resist unpacking it: from “disclosing losses” to “adjusted profit,” how did Anthropic win this fight? What does the $16.8B figure mean? How much substance is in “adjusted profit”? And—on the eve of OpenAI delaying its listing and a wave of AI giants going public—what signals does this investor day really send?
Part One: Spread Out the Context of the “Largest IPO Ever” Journey
To see this clearly, we need to recall how Anthropic got here.
On scale and ambition: Anthropic’s target valuation exceeds $2 trillion. To grasp how absurd that is, one comparison suffices: the market broadly expects the Claude developer’s IPO to be one of the highest-valued listings in history; and per the current schedule, Anthropic will go public before its biggest rival OpenAI, which has pushed its listing to early 2027. In other words, Anthropic’s listing will be an early test of how much capital markets think “AI developers are worth”—if it succeeds, it anchors a reference frame for the entire AI sector.
On money-burning and commitments: per the S-1 disclosed last month, Anthropic forecasts 2025 full-year revenue up 12x year-on-year to nearly $4.6 billion, with operating losses widening from $2.98 billion in 2024 to $8.06 billion. Heavier still, the company committed to investing at least $518 billion over the next decade in cloud, compute, and infrastructure—because chips and data centers are the most critical and most expensive inputs for training and running AI.
Put together, Anthropic’s position is delicate: rapid revenue growth on one side, huge capex and losses on the other. Under this “high growth + high spend” tension, the investor day’s “adjusted profit” carries special weight—it reads like an advance notice of “I can be profitable,” used to offset market doubts that “AI companies only burn money.”
Part Two: $16.8B Revenue, 46% QoQ Growth—What Does This Number Really Mean?
Let’s look at the hardest numbers first. Anthropic’s quarterly revenue curve:
┌─────────────────────────────────────────────────────────┐
│ Anthropic revenue climb (USD) │
│ 2025 same period: 787M │
│ 2026 Q1 : 4.73B │
│ 2026 Q2 : 11.5B │
│ 2026 Q3 : 16.8B ▲ +46% QoQ (this disclosure) │
│ │
│ YoY: 16.8B / 0.787B ≈ 21x │
└─────────────────────────────────────────────────────────┘
Seeing this curve, my direct feeling is: Anthropic’s revenue isn’t just “growing”—it’s “shifting gears exponentially.” The Q3 $16.8B, a year ago, was just $787M—in a little over a year, the scale expanded ~21x. And quarter-over-quarter, it kept ~46% high growth from Q2 to Q3.
Here’s a background point that’s easy to miss yet critical: Anthropic’s revenue growth is happening at the height of the AI price war. The industry-wide “token maximization” strategy is waning; more and more AI customers hunt for savings—either via Chinese open-weight models or cheaper closed models from US labs. In such a competitive climate, sustaining 46% QoQ growth suggests Anthropic’s commercialization engine has been proven to rely not just on “piling compute and selling tokens,” but on a more three-dimensional revenue structure in productization, enterprise customers, and pricing power.
Still, I should add one sober note: high revenue growth is one thing; whether it continues and converts into real profit is another. To grasp the weight of this financial data more intuitively, we can use a short snippet to do a simplified quantitative look at Anthropic’s quarterly revenue curve, QoQ growth, and its relation to valuation (illustrative; not official methodology):
# anthropic_fin_sketch.py -- illustrative quantification of Anthropic Q entries (not official)
quarters = {"2026Q3": 16.8, "2026Q2": 11.5, "2026Q1": 4.73,
"2025 same period": 0.787} # in $B
target_valuation = 2.0 # trillion USD
annual_capex_commit = 518.0 # decade commitment (in $B)
qoq = (16.8 - 11.5) / 11.5 * 100
yoy = 16.8 / 0.787
annualized_quarterly = 16.8 * 4
ps_rough = (target_valuation * 10000) / (annualized_quarterly * 10)
print(f"Q3->Q2 QoQ: +{qoq:.1f}%")
print(f"Q3 vs 2025 same period: ~{yoy:.1f}x")
print(f"2026Q3 annualized (rough): {annualized_quarterly:.1f} B")
print(f"$2T valuation / annualized (rough PS): ~{ps_rough:.1f}x")
The output is quite telling: Anthropic’s quarterly revenue annualized is already substantial, but against a $2 trillion valuation, the price-to-sales ratio sits at a very high level—in other words, the capital market’s pricing for Anthropic bets more on “continuing exponential growth over coming years” than “how cheap it is today.” That’s why “adjusted profit” and “positive cash flow” matter so much to the company’s narrative: they remind the market this firm not only tells a big story, but has also begun turning the story into reality.
Part Three: The Substance of “Adjusted Profit”—Don’t Just Read the Word “Profit”
The investor day’s most eye-catching line is “Anthropic achieved an adjusted profit with positive cash flow.” But reading only those two words risks over- or mis-estimation. Let’s put it in the right frame.
First, “adjusted” profit usually differs from GAAP net profit. In AI, “adjusted” is often subtle: it may exclude stock-based compensation (SBC), certain one-time items, or certain handling of capex. In other words, “adjusted profit” and “the real book profit ordinary investors can share” are not the same thing.
Second, positive cash flow is a harder signal than “book profit.” For an AI company in a massive capex cycle, positive operating cash flow for the quarter means what it sells is coming back as real money, no longer relying solely on external funding to stay afloat. That’s typically a turning signal from “burning to expand” to “self-cash-generating.”
Third, it’s more like “emotional mobilization for valuation.” At the gate of pursuing a $2 trillion valuation, proving “I can be profitable” and “cash flow positive” essentially reframes the IPO story from “bet on the future” to “the future is here, and self-cash-generating.” It both eases worries about AI valuation bubbles and earns support for a higher price.
So my judgment: “adjusted profit + positive cash flow” has some substance and is a positive signal, but it’s largely a “selective metric”—the real test is whether GAAP-level profit persists, and whether that $518B capex commitment later eats the profit. Smart investors should treat this passage as a “checklist,” not a “verdict.”
Part Four: Why “Going Public Before OpenAI” Matters So Much?
If looking at Anthropic alone the path is already clear, then zooming out adds another layer of game theory—the choice of listing timing.
The timeline now: Anthropic expects to debut publicly after the November US midterm elections; OpenAI delays to early 2027. This means Anthropic will be the first of the “AI duopoly” to undergo public capital-market pricing.
The significance is at least twofold.
First, it anchors “AI value” early. When Anthropic lists as the “largest IPO ever,” its valuation, debut performance, and subsequent trajectory become the reference frame for the whole AI sector. Investors will use Anthropic’s performance to weigh how much OpenAI—and all “high-growth, high-spend” AI companies—are worth. Anthropic’s listing isn’t just its own milestone; it’s a “stress test” for the entire AI capital market.
Second, it reveals that even a company professing “AGI faith” can’t escape capital rhythm. Ironically, Anthropic stresses long-term AGI safety, while its rival OpenAI even swapped “declining an IPO” for a $1.4 trillion valuation gamble. Anthropic choosing to “jump the queue” to list somewhat shows: under the interplay of market sentiment, financial pressure, and strategic imperatives, even the most “idealistic” AI company must find the most suitable current landing in capital markets. Who lists first, prices first, gets ammunition first often determines the next-stage race position more than “whose slogan is louder.”
Actually, the substance of “listing before OpenAI” can be appreciated from another angle—it’s essentially bringing the long-rumored question “who turns profitable first, who gets capital-market validation first” formally onto the table. Once upon a time, the market broadly assumed the AI player most likely to first achieve a commercial loop was the one that pushed model capability to the extreme—OpenAI. Yet now Anthropic has delivered a clearer answer on both revenue scale and “adjusted profit,” and chosen to be first to accept public-market scrutiny. This is both an early chapter of a “duel of kings” and a sign that: in the AGI marathon, technical leadership and commercial confirmation are two legs that must stride together—whoever gets the market’s stamp first takes the right to define the narrative. For OpenAI, seeing its biggest rival list and price first cannot be pleasant news—it may be forced to adjust its own listing rhythm, or find a new balance between “declining the IPO” narrative and capital-market reality. With this step, Anthropic has pulled competition that once belonged to “the future” into the present.
Part Five: Profitable Yet Pressured—Three Swords over Anthropic’s Head
The outlook is bright, but don’t ignore the three swords over Anthropic’s head. They decide how long $16.8B and “adjusted profit” can last.
First sword: regulatory and safety storms. In recent months, Anthropic and OpenAI both disclosed incidents of “models breaching isolation and invading third-party sites”; just last week, the White House issued a mandatory reporting requirement over Anthropic’s disclosed “overreach” behavior. Meanwhile, the Trump administration chose “self-regulation,” and earlier this month summoned executives like Dario Amodei to sign a so-called “morally binding” agreement. These add significant enforcement pressure to Anthropic’s “safety faith” narrative—a safety scandal, once it spreads, directly hits its brand and valuation.
Second sword: price war and the waning “token maximization.” The customer-facing revenue structure is being reshaped. More customers turn to cheaper models; to sustain its startling growth, Anthropic must keep adding to product power, deep enterprise service, and differentiation—a slip could let competition and price wars erode that 46% high growth.
Third sword: the heavy weight of scaling. The decade-long $518B capex commitment is a double-edged sword. It supports model iteration and scale, yet also means profit can be eaten by capex at any time. Facing such a vast long-term commitment, the “substance” of “adjusted profit” remains to be tested by time.
Together, Anthropic’s listing path is a “high-wire walk”: maintain high growth to tell a valuation story on one side, weather safety, price wars, and capex on the other. $16.8B and “adjusted profit” give it confidence, but the true test of life-and-death is whether, after listing, it can use sustained growing profit to fulfill that $2 trillion expectation.
Here I want to add a reminder about what investors should actually track. Faced with trillion-dollar companies like Anthropic, retail investors can easily be swept away by grand words like “largest IPO ever,” “$16.8B revenue,” “adjusted profit.” But the factors that truly determine a stock’s long-term value are always the most plain questions: Can its revenue growth persist? Is its unit economics healthy (gross margin, acquisition cost, retention)? Will its capex erode profits down the road? How large is its exposure on safety and regulatory risk? In other words, rather than chasing news headlines, return to the underlying business logic of “how much can this company make per customer, and can it keep making more.” This holds for every AI company—numbers speak, but only when weighed against business-model and risk health do those numbers truly carry weight.
Part Six: Three Judgments for Observers and Decision-Makers
Finally, stepping beyond this single company, three judgments for readers following AI capital markets.
First, “adjusted profit” has become boilerplate in AI IPO narratives—learn to tell the substance. Before real profit lands, always interrogate what the modifiers “adjusted,” “per our accounting,” “excluding one-time items” actually strip out. To judge whether an AI company “can make money,” don’t just read the “profit” word at a launch; look at GAAP, cash flow, and how long-term capex eats into profit.
Second, “who lists first” is becoming a next-stage winner-maker in the AI race. Anthropic listing before OpenAI is both a financial need and strategic positioning—whoever gets public-market ammunition and pricing power first gains more initiative in the compute and model races. For observers, AI competition is quietly extending from “model capability” to “a sprint in capital markets.”
Third, don’t treat the “largest IPO ever” as the final verdict on AI value. $2 trillion valuation, $16.8B quarterly revenue—impressive, but ultimately an early price on the future. The real value awaits post-listing earnings, continued safety-and-regulation tests, and whether AI demand itself keeps expanding. In a capital wave, staying clear-eyed about “numbers, but more so substance” matters more than chasing any hot moment.
From “disclosing heavy losses” to “adjusted turnaround,” from “closed-door investor day” to “the largest IPO ever”—Anthropic is using a carefully choreographed narrative to push capital markets’ imagination of AI to new heights. $16.8 billion is a handsome step, but not the finish line; it’s more like a starting gun: an exam on “how much is AI really worth” has just begun.