Everything About AI
NextFin News - Anthropic filed to go public on Monday with a $16.3 billion first-half revenue run that is 14 times a year earlier and an $80.7 billion net loss that is almost entirely a paper accounting charge - while HSBC reported a 32% jump in third-quarter pretax profit to $9.6 billion that still failed to satisfy investors because the engine underneath, its wealth business, is stalling. The two disclosures, hours apart on opposite sides of the world, frame the central tension of this earnings season: the AI boom is printing revenue at a pace no one predicted, but the accounting reality of funding it is brutal, and the old financial order is growing again without convincing anyone it can grow fast enough.
The Claude maker's Form S-1, filed with the Securities and Exchange Commission on October 26 and set to list on the Nasdaq Global Select Market under the ticker ANT, is the most revealing look yet at what a frontier AI laboratory actually looks like as a public company. It shows a business that turned a $974 million operating profit in the second quarter - its first profitable operating quarter in the six reported - while the consolidated entity booked an $80.7 billion loss because the fair value of its convertible notes, derivative liabilities and SAFEs rose along with the company's own valuation. In other words, the more valuable Anthropic became, the bigger the loss it had to report. That is not a cash burn story. It is a valuation-feedback story, and it is about to be put in front of public-market investors who price every quarter.
HSBC's problem is the mirror image. Reported profit before tax rose 32% to $9.6 billion, revenue climbed 5% to $18.6 billion, net interest income gained 6% to $9.3 billion, and operating expenses fell 12% to $8.8 billion. Yet the headline number carried a $1.4 billion favorable swing from notable items, largely the non-recurrence of legal provisions, and revenue of $18.6 billion came in just under the roughly $19.0 billion consensus. The bank's shares slipped as investors focused on the slowdown in the wealth franchise that management has been betting its future on. A bank that looks cheap on reported earnings can still look expensive if its growth engine is losing momentum. A startup that looks unprofitable on a headline loss can still look cheap if its operating profit has already turned. The market is being asked to decide which distortion matters more.
The $80.7 Billion Loss That Is Not What It Looks Like
The first question every investor will ask about the Anthropic filing is the simplest one: how can a company that is supposedly worth close to $1 trillion lose $80.7 billion in six months? The answer sits on page 23 of the S-1. Of the $80.7 billion net loss in the first half of 2026, $72.4 billion - 89.6% of the total - came from non-cash fair-value changes on the convertible notes, derivative liabilities and SAFEs the company issued to raise funding. Those instruments increase in value as Anthropic's equity value rises, and the company must mark that increase as an expense even though no cash leaves the building. Compare that with the $693 million operating loss from running the actual business over the same period, and the picture changes entirely: the reported loss is a valuation-feedback loop, not an operating collapse.
The revenue tells the other half of the story. First-half revenue reached $16.29 billion, up from $1.16 billion a year earlier - a 14-fold increase. In the second quarter alone, the company earned $974 million of operating income, its first profitable operating quarter in the six quarters shown in the filing. That is the inflection point the market will trade on: a company whose core operations have already crossed into operating profitability while its reported net income remains grotesquely negative because of accounting rules designed for a different kind of capital structure.
The catch is that the accounting is not the only thing getting bigger. The S-1 lays bare the scale of the compute commitments Anthropic has signed with its backers. Minimum spending owed to Amazon and Google runs to at least $224.5 billion over contracts extending to 2033 and 2036; including Microsoft, the total reaches $269.4 billion. The SpaceX arrangement alone commits up to $120.5 billion through 2029, at roughly $3.1 billion a month once all capacity is delivered, which the company expects in December 2026. Those are fixed obligations that survive any demand slowdown. They are the price of staying in the frontier race, and they are also the reason the operating-profit turn may not translate into free cash flow for years.
There is a second-order implication here that the market has not fully priced. Anthropic and OpenAI together now account for roughly half of the revenue backlog across the four largest U.S. cloud providers - commitments that the cloud giants are booking as future revenue and that investors are already valuing into their shares. If Anthropic's demand growth slows, the pain does not stop at ANT. It transmits directly to Amazon, Microsoft, Google and the chip supply chain, because the backlog that supports their growth narrative is concentrated in a handful of AI labs whose spending is funded by paper valuations rather than durable cash flow. The IPO does not just test appetite for one stock. It tests whether the entire cloud-growth trade is built on committed demand or on financing that could evaporate in a risk-off episode.
"We believe that frontier AI companies must invest in safety and alignment to help ensure the responsible diffusion of this technology," Anthropic said in its prospectus.
That sentence matters more than it reads at first glance. Anthropic is a public benefit corporation, and in October it announced it would commit 5% of its shares to a public benefit fund. The filing explicitly flags the risk that its models could develop capabilities faster than the company can keep them safe, or that they could be used to cause harm. This is a company asking public shareholders to accept a dual mandate - profit and public benefit - while it spends hundreds of billions on infrastructure and warns that its own product could pose an existential risk. That is not a standard tech IPO. It is a social contract wrapped in a growth equity.
HSBC's Profit Jump and the Growth Engine That Is Not Firing
HSBC's third-quarter numbers look strong until you ask what produced them. Reported profit before tax rose 32% to $9.6 billion from $7.3 billion a year earlier, and revenue grew 5% to $18.6 billion. But $1.4 billion of that came from a favorable swing in notable items, driven mainly by the non-recurrence of legal provisions. Strip that out and the underlying momentum is thinner. Net interest income rose 6% to $9.3 billion and operating expenses fell 12% to $8.8 billion, helped by cost-simplification savings that Group CEO Georges Elhedery has upgraded from $1.5 billion to $2 billion to be actioned before the end of this year and delivered in full in 2027.
The problem is where the growth is not. The bank's wealth franchise - the business HSBC has positioned as its crown jewel and its Asia-centric future - slowed markedly in the quarter. Wealth management is supposed to be the high-margin, fee-driven engine that reduces dependence on interest rates, and it is the area where investors were looking for acceleration. Instead they got deceleration. The market reaction was immediate: despite the headline strength on reported profit, the shares drifted lower as the focus shifted from what the bank earned to how sustainably it can earn it.
This is a cyclical problem layered on a structural bet. The cyclical leg is straightforward: wealth flows slow when markets wobble and when high interest rates keep cash competitive with fee-based products. That part can revert if rates fall and equity markets stabilize. The structural leg is more uncomfortable: HSBC is competing for Asian wealth against private banks, local champions and the digital platforms that are compressing fees across the industry. Cost-cutting can defend margins for a while - the upgraded $2 billion savings target proves management knows this - but it cannot manufacture the top-line growth that a wealth franchise needs to justify its valuation multiple.
There is also a distribution question the market will keep asking. HSBC announced a third interim dividend of $0.10 per share and a new share buyback of up to $1 billion, after completing an earlier $1 billion program that repurchased approximately 48.7 million shares. Capital return is generous, and it is partly why the stock has support. But a bank that returns capital while its growth engine slows is signaling that it cannot deploy that capital at attractive returns internally. Buybacks are a floor under the shares, not a substitute for the revenue growth that would re-rate them.
The two stories converge on one point: reported earnings are becoming less informative than the structure underneath them. At Anthropic, the reported loss hides an operating turn. At HSBC, the reported profit hides a growth stall. In both cases the headline number is the wrong place to start.
What the Market Is Pricing In - and What It Is Missing
The market reaction on October 26 showed which narrative investors preferred. The Dow Jones Industrial Average rose 0.28% to 52,361.76, the S&P 500 gained 0.70% to 7,677.03, and the Nasdaq Composite advanced 0.83% to 26,169.38, with semiconductor shares leading as investors assessed the prospects for AI-related spending. The filing lifted AI sentiment across the board, reversing two consecutive weeks of losses. That is a market choosing to believe the operating-profit inflection story and to treat the $80.7 billion loss as an accounting artifact.
But the valuation math is unforgiving. Anthropic's last private round, a $65 billion Series H in May, valued the company at $965 billion. Reports ahead of the filing have floated a potential listing valuation of up to $2 trillion and a raise of more than $60 billion, which would make it one of the largest IPOs in history. At $965 billion, the company trades at roughly 30 times its annualized first-half revenue run-rate; at $2 trillion, that multiple nearly doubles. For that to be justified, the operating-profit turn has to scale into free cash flow fast enough to outrun the $224.5 billion-plus compute bill. The window between "profitable on an operating basis" and "cash-generative after capex and compute commitments" is where most AI bulls are doing their most optimistic work.
Not everyone is convinced. "Strong AI demand alone may no longer be enough," Charu Chanana, chief investment strategist at Saxo, said ahead of the earnings week. "Markets want evidence of monetization, margins and cash flow." That is the counter-thesis in one sentence: revenue growth at 14 times the prior year is impressive, but the market has moved past growth-at-any-cost and is now demanding proof that the growth converts into cash after the infrastructure bill is paid. Anthropic's operating profit in the second quarter is the first piece of that proof. The compute commitments are the reason it is not yet the whole proof.
The Adversarial Case: Why the AI Rally Could Roll Over Anyway
The strongest argument against buying the Anthropic-led AI rally is not about the company's fundamentals. It is about positioning, financing and the mechanics of a crowded trade. The AI trade has been crowded for years, and the cloud backlogs that support the megacap growth narrative are now heavily concentrated in a small number of AI labs whose spending is funded by private capital raised at stratospheric valuations. Anthropic and OpenAI alone account for roughly half of the revenue backlog across the four largest U.S. cloud providers. When so much of the market's growth story rests on the spending plans of two privately funded companies, the risk is not just that one company misses - it is that a valuation reset in one propagates through the whole chain.
If the IPO prices at the high end of the reported range and then trades down - the familiar pattern for hyped debuts, where early enthusiasm gives way to quarterly scrutiny - the feedback loop runs in reverse. Lower public valuations mean larger fair-value losses on the very same convertible instruments and SAFEs, which means uglier reported net losses, which means more pressure on the stock, which means tighter and more expensive financing for the whole AI cohort. The accounting quirk that makes Anthropic look worse than it operates can work in both directions, and in a risk-off episode it becomes a multiplier rather than a footnote.
There is also a demand-side risk that the filing itself highlights. Anthropic's compute commitments are fixed obligations; its revenue is not. If enterprise AI spending growth slows - if the 43.8% of U.S. businesses paying for Anthropic services in September, per market intelligence firm Ramp, stops expanding - the company is left paying for capacity it cannot monetize at prior margins. The industry's own leaders are signaling caution: OpenAI's CEO has called a listing this year "ill-advised" amid AI safety concerns, and Anthropic's CEO, Dario Amodei, has argued for slowing the development of the most advanced models. When the people building the technology are publicly debating whether to slow down, the assumption of uninterrupted hypergrowth deserves a discount.
The historical precedent is not reassuring. Mega-cap technology debuts that price at the top of their range tend to underperform over the following year as the market transitions from pricing the story to pricing the quarterly execution. Anthropic will face that transition with a fixed cost base that keeps growing regardless of demand. A bull must believe that revenue growth outruns a $224.5 billion-plus compute bill; a bear only needs to believe that growth decelerates once. That asymmetry favors the skeptic until the first few public quarters prove otherwise.
The falsifying signal for the bullish read is concrete: if Anthropic's first public quarter shows operating income shrinking sequentially while compute expense keeps rising, or if the company discloses that its cloud minimums are being restructured because demand cannot absorb the capacity, the inflection thesis breaks. Watch the operating-margin trajectory and any change to the $224.5 billion commitment figure. A stable or rising commitment with falling operating leverage is the exact pattern that would prove the bear case right.
What Comes Next: Scenarios Across Three Time Horizons
In the short term, sentiment is with the AI trade. The filing revived a market that had lost two straight weeks, and the IPO itself - likely to price in November at the earliest, once the SEC review and roadshow process runs its course - will be a liquidity event that keeps the sector in focus. HSBC, by contrast, faces a near-term overhang: investors will press management on the wealth slowdown at the analyst meeting, and until there is evidence of reacceleration, the shares are likely to trade on capital-return support rather than multiple expansion.
Over the medium term, the divergence widens. The base case for Anthropic is a successful listing at a valuation between the last private round and the reported aspirational ceiling, with operating profitability expanding but free cash flow remaining negative as compute spending ramps through 2027. The upside case is that agent-based products - the Cowork and Duet offerings that have already rattled software stocks - convert usage into durable enterprise contracts faster than expected, pushing revenue toward the company's internal projections and justifying a premium multiple. The downside case is a priced-at-the-top debut followed by a post-lockup reality check, where the fixed compute bill collides with softer demand growth and the stock rerates toward more conventional software multiples.
For HSBC, the base case is steady capital returns and low-single-digit revenue growth, with the stock held up by the buyback and the $0.10 quarterly dividend but capped by the wealth slowdown. The upside case is that Asian wealth flows reaccelerate as rates fall and regional equity markets recover, turning the franchise back into the growth engine management has promised. The downside case is that fee compression and competition in Asian wealth prove structural rather than cyclical, forcing another round of cost cuts and dividend caution.
Long term, the structural call is this: Anthropic's operating-profit turn is real, but whether it compounds depends on demand growing faster than the fixed compute bill - a race the company is currently winning but cannot take for granted. HSBC's profit jump is real, but the wealth franchise's slowdown is the more durable signal unless Asia's wealth cycle turns. One is a growth story with a financing problem; the other is a value story with a growth problem. The market has voted for the first and discounted the second.
The kicker for investors is that both distortions will resolve the same way: through cash flow. Anthropic must prove its operating profit survives the compute bill. HSBC must prove its wealth engine survives the rate cycle. Until then, the reported numbers - the $80.7 billion loss and the 32% profit jump - are just the noise between the signal.
Explore more exclusive insights at nextfin.ai.
