Alibaba Is Up 70% on AI. Investors Still Don't Own What They Think They Own.

---

Let us begin by taking the bull case seriously, because it is genuinely strong and a forensic piece that strawmans it would be worthless. Alibaba in 2026 is not the wounded, politically marked company of a few years ago. Its cloud division has gone from 18% growth to 26% to 34% and most recently to roughly 38%, reaching around $6 billion in quarterly revenue, with external cloud revenue accelerating to 40% and AI-related products now making up about 30% of external cloud sales — an eleventh consecutive quarter of triple-digit AI growth. Its Qwen family of models is among the most widely adopted open-source AI ecosystems in the world. It reported fiscal 2026 net income of roughly $15 billion. It sits on about $41 billion of net cash. And it trades at something like 22 times earnings — a fraction of what comparable American cloud-and-AI companies command. On every operational metric, this is a world-class technology business re-accelerating into the biggest trend of the decade, available at a discount that looks, on its face, absurd.

The bulls have a one-word explanation for the discount: temporary. The China crackdown is over, they say; Beijing has pivoted from punishing its tech champions to enlisting them in the AI race against America; Xi shook Jack Ma's hand; the regulatory winter has thawed; and so the gap between Alibaba's valuation and its American peers' will close as the fear recedes and the world remembers what a great business this is.

This essay is about why that explanation is half right and dangerously incomplete — about what the discount actually compensates for, why it is structural rather than temporary, and why the very gesture the bulls cite as proof the risk has passed is in fact the clearest evidence of what the risk has always been.

The footnote that owns the company

Start with the structure, because it is the single most important and least discussed fact about every Chinese company listed in the United States, Alibaba included. When you buy a share of "Alibaba" on the New York Stock Exchange, you are not buying a share of the Chinese company that runs Taobao, Tmall, and Alibaba Cloud. You are buying a share of a holding company incorporated in the Cayman Islands. That Cayman shell does not own the Chinese operating businesses either — Chinese law restricts or forbids foreign ownership of those sectors. Instead, the shell controls the operating businesses through a structure called a variable interest entity, or VIE: a stack of contracts that attempt to route the economic benefits of the Chinese companies up to the foreign holding company without conveying actual equity ownership.

Read that again, slowly, because it is the whole game. The foreign shareholder owns a contract that promises the profits of a business, not the business. And here is the detail that the company itself discloses and the rally ignores: those VIE contracts have never been tested and validated in a Chinese court. Their enforceability — the entire legal foundation on which hundreds of billions of dollars of foreign investment in Chinese tech rests — is an assumption, not an adjudicated fact. If a Chinese court, at the direction of the Chinese state, were ever to rule those contracts unenforceable, the foreign shareholders' claim on the underlying business could, in principle, evaporate. This is not a fringe theory; it is the standard risk disclosure in the 20-F of every major U.S.-listed Chinese company, and it is the reason the discount exists.

The bull says this risk is theoretical, that the VIE structure has functioned for two decades, that Beijing has every incentive to keep foreign capital flowing and would never blow up the structure. All of that is true, and none of it changes the fact that the structure rests on the continued forbearance of a party-state that does not answer to shareholders. A risk that has not materialized is not a risk that does not exist. It is a risk that has not materialized yet.

The crackdown was the proof, not the aberration

To understand why the discount is structural, you have to remember what happened the last time the world decided Chinese tech was a one-way bet — because it was not long ago, and it was not subtle.

In late 2020, Alibaba's affiliate Ant Group was days from what would have been the largest initial public offering in history. Beijing killed it, abruptly, after Jack Ma gave a speech mildly critical of Chinese financial regulators. Ma, the most famous entrepreneur in China, then vanished from public view for months. Over the following two years, the Chinese state launched a sweeping "rectification" of its technology sector — antitrust fines, data-security crackdowns, the effective destruction of the private tutoring industry overnight, new restrictions across gaming, fintech, and e-commerce. Something on the order of a trillion dollars of market value was erased from Chinese tech stocks. Alibaba itself fell roughly 80% from its peak. None of it was driven by deteriorating business fundamentals. It was driven by the state deciding that its technology champions had grown too powerful, too independent, too rich, and needed to be reminded who was in charge.

That episode is the single most important data point for valuing Alibaba, and the bull case treats it as ancient history. It is not history. It is the demonstration — the proof of concept — of exactly what the VIE discount prices: that in China, shareholder value is subordinate to state objectives, and the state will, without warning and without regard to investors, destroy enormous amounts of that value when its political calculus changes. The crackdown was not an aberration from which China has now recovered. It was the system revealing its actual priorities. A market that watched a trillion dollars vanish on political command and then concluded, a few years later, that the risk was "temporary" has a short memory.

The handshake is the risk, not its absence

Now to the gesture the bulls cite most often as evidence the danger has passed: the public rehabilitation of Jack Ma, the meetings between Xi Jinping and China's tech founders, the state's evident decision to enlist Alibaba as a pillar of national AI ambition. The rally reads this as the all-clear — Beijing loves its tech champions again, so the discount should close.

Look at the same facts through a forensic lens and they say something almost opposite. The fact that Alibaba's fortunes can swing so violently on whether the paramount leader chooses to shake its founder's hand is not evidence that the political risk has gone away. It is the political risk, displayed in its purest form. When the single most important variable for a half-trillion-dollar company is the personal disposition of the head of a one-party state toward its founder, you are not looking at a normal equity. You are looking at an asset whose value is a function of political favor — and political favor, by its nature, is granted conditionally and can be withdrawn at will. The handshake that lifted the stock is the same mechanism, pointed the other way, that the killed Ant IPO represented. A company that rises on a handshake is a company that can fall on a frown. The volatility of the favor is the proof of the dependence on it.

This is the deepest reason the discount is structural. It does not compensate for a specific, resolvable problem that good news can cure. It compensates for the permanent condition of operating at the pleasure of a state that has both demonstrated its willingness to destroy shareholder value and constructed a legal structure (the untested VIE) that would let it do so with deniability. No earnings beat, no cloud-growth acceleration, no Qwen benchmark can resolve that condition, because the condition is not about the business. It is about who ultimately controls the business, and the answer is not the shareholders.

The delisting tail

Layered atop the VIE risk is a second, more mechanical threat that the discount also prices: the possibility of forced delisting from U.S. exchanges. Under the Holding Foreign Companies Accountable Act, a Chinese company can be delisted from American markets if the U.S. accounting regulator, the PCAOB, loses inspection access to its auditors for two consecutive years — a function of the long-running standoff between Washington and Beijing over audit oversight.

In fairness, this is genuinely a tail risk rather than a base case. The PCAOB currently retains inspection access, and Alibaba has prudently established dual primary listings in both New York and Hong Kong, so that if its American depositary shares were ever delisted, holders could in principle convert into Hong Kong–listed shares and most of the value would migrate rather than vanish. The Hong Kong backstop meaningfully reduces the severity of the delisting scenario, and a fair analysis must credit it.

But notice what even the mitigated version requires the American investor to accept: that their holding could, on a geopolitical breakdown beyond their control or the company's, be forcibly converted out of the market and the currency they bought it in, into a Hong Kong listing subject to Chinese jurisdiction. The backstop softens the blow; it does not eliminate the fact that a U.S.-listed Chinese stock is a hostage to the U.S.–China relationship, exposed to a deterioration that has nothing to do with Alibaba's cloud growth and everything to do with forces in Washington and Beijing that no shareholder can hedge. That exposure, too, is part of what the discount is paying for — and it is not temporary, because the geopolitical rivalry that produces it is the defining structural feature of the era.

The AI re-rating is being bought with profit

There is one more thread, and it is the operational one, because even the pure business story has a complication the rally underplays. Alibaba's AI and cloud re-acceleration is real — but it is being purchased at a steep and rising cost to profitability. The company's capital expenditure has surged, with fiscal 2026 capex reaching roughly RMB126 billion, up from about RMB84 billion the prior year, and management has signaled it will likely overshoot even its already-enormous multi-year AI commitment (north of $50 billion over three years) as data-center needs explode. The bottom-line cost has been severe: group non-GAAP net income fell roughly 62% for the fiscal year as the spending ramped, and in the March 2026 quarter adjusted net income collapsed almost entirely — the company posted its first operating loss since the COVID era — even as the top line accelerated.

This is the same dynamic playing out at every hyperscaler — spend enormously now to capture AI demand later — but it carries a particular irony at Alibaba, because the cheap multiple that anchors the bull case is partly a function of the profit the company is now spending down. If you are buying Alibaba at "22 times earnings," you are buying it at 22 times an earnings number that the AI build-out is actively pressuring. The re-rating story ("the AI champion deserves a higher multiple") and the value story ("it's cheap on current earnings") are in quiet tension: the AI spending that justifies the former erodes the earnings that produce the latter. That does not make the investment bad. It makes the "cheap" framing less clean than the headline 22-times suggests.

It isn't an Alibaba problem; it's a China-structure tax

A useful way to confirm that the discount is structural rather than company-specific is to look sideways at Alibaba's peers, because if the discount were really about Alibaba's particular business, it would not apply uniformly to every other U.S.-listed Chinese champion. It does. PDD Holdings, the parent of Pinduoduo and Temu, grows faster than almost any large retailer on Earth and still trades at a China-ADR discount. JD.com, a logistics and e-commerce operation of staggering scale, trades at a discount. The entire basket of Chinese internet stocks — the kind tracked by the KWEB ETF — carries the same risk premium, applied across wildly different businesses with wildly different fundamentals. When a discount attaches not to a company but to a jurisdiction and a legal structure, that is the definition of a structural, non-diversifiable risk. You cannot stock-pick your way out of it, because it is not about the stock.

This is why the "Alibaba is uniquely cheap" framing slightly misleads. Alibaba is cheap for the same reason PDD and JD are cheap: because all of them are VIE-structured, state-exposed, delisting-vulnerable claims on Chinese businesses, and the market applies a consistent haircut to that entire category. The haircut can narrow when sentiment warms and widen when it cools — it has done both, violently, over the past five years — but it does not go to zero, because the underlying conditions that justify it do not go to zero. An investor who buys Alibaba expecting the discount to fully close is implicitly betting that the market will one day decide that owning a contract over a Chinese asset is exactly as safe as owning an American company outright. That day is not coming, and the steadiness of the peer-wide discount across every Chinese ADR is the evidence.

There is even a subtle tell in Alibaba's own capital allocation. A company sitting on $41 billion of net cash, returning capital through buybacks and dividends rather than deploying every yuan into its booming opportunities, is in part a company acknowledging that there are limits — political, regulatory, strategic — on how freely it can grow inside its own system. Capital returns are shareholder-friendly and welcome, but at a 40%-growth cloud company they also quietly signal that not every dollar can find a politically safe home in further expansion.

What the bulls genuinely get right

In fairness, and it matters: the bulls may still make money, and the operational case is not a mirage. Alibaba is a genuinely excellent business — dominant Chinese e-commerce, a world-class cloud platform, a frontier AI model family, and a fortress balance sheet with $41 billion of net cash that very few companies on Earth can match. If the VIE structure continues to function as it has for twenty years, if Beijing continues to favor its AI champions, if the U.S.–China relationship avoids catastrophic rupture, and if the cloud growth keeps compounding at 40%, then Alibaba at 22 times earnings is genuinely cheap and the discount genuinely does close, and the patient holder is rewarded handsomely. That is a real and reasonable scenario, and the stock's 70% recovery reflects a market beginning to price it.

The honest synthesis is not "Alibaba is a bad investment." It is that the discount the bulls dismiss as temporary is in fact the market's accurate, permanent pricing of a set of risks — VIE non-ownership, state subordination of shareholders, delisting exposure — that no business achievement can dissolve, because they live above the business, in the realm of law and politics where shareholders have no vote. The discount is not a mistake the market is about to correct. It is the price of a structural truth, and the question for any buyer is not "why is it so cheap" but "am I being paid enough for the thing I am actually buying" — which is a contract, in a shell, in a jurisdiction where the state has shown exactly what it will do.

The kicker

The seduction of the Alibaba trade is that it lets you buy a world-class AI company at a value-stock price, and the seduction is real because the company is real and the price is low. But the discount is not a glitch waiting to be arbitraged away by a market that finally "gets it." The market gets it. The discount is the considered judgment that a Cayman shell holding untested contracts over Chinese assets, at the sufferance of a state that once erased a trillion dollars to make a point, is worth less per dollar of earnings than a company you actually own in a jurisdiction with enforceable property rights. The bulls call that gap an opportunity. The filings call it a risk. Both can be right for years — until the day they are not, and the thing that was always true becomes the only thing that matters.

You can own the growth, the cloud, the models, the cash — or rather, you can own a contract that promises them, in a shell, an ocean away, validated by no court. Somewhere in Hangzhou a business compounds at forty percent a year, and somewhere in Beijing a decision sits unmade, and the distance between those two places is the entire discount, and it has never once been temporary.