Scoreboard

Hey there, bargain hunter. Alibaba just announced the most aggressive capital raise in Hong Kong’s history, and the market’s reaction tells you everything about where the debate sits right now.

On August 23, 2026, Alibaba Group (NYSE: BABA / HKEX: 9988) launched a placement of 710 million ordinary shares at HK$112.70 each, targeting HK$80 billion, or roughly $10.2 billion. That price represented a 3.6% discount to Friday’s close. Within hours of the announcement, Michael Burry went public with his exit, declaring he had swapped his BABA position for a large stake in JD.com, and that Alibaba’s shares would need to fall 50% before he would consider returning. BABA shares fell 8.57% on the news.

That is the scoreboard: a record-breaking share sale, a famous skeptic walking out the door, and a stock sliding on a day when the company claimed its fastest cloud growth in 22 quarters. Two investors can look at identical facts and reach opposite conclusions. That is exactly what is happening here.

The Real Reason This Deal Happened

Alibaba is not raising capital because it is flush. It is raising capital because its three-year AI spending commitment is burning through cash faster than its core commerce business can replenish it.

In February 2025, CEO Eddie Wu announced a plan to invest at least RMB 380 billion ($53 billion) over three years on AI and cloud infrastructure. That figure already exceeded Alibaba’s total AI and cloud spending over the prior decade. By August 2026, Alibaba had already spent roughly half that commitment. Quarterly capital expenditure jumped 75% year over year to 68 billion yuan in the June quarter alone. The result: free cash outflow of $6.6 billion and a net profit collapse of 75% to 10.5 billion yuan ($1.6 billion), even as revenue grew 9% to 269 billion yuan.

This is the fundamental tension bargain hunters need to hold in their heads simultaneously. Revenue is growing. Cloud is accelerating hard. And earnings are being intentionally destroyed by the pace of the buildout. That is not a business in trouble. It is a business making a deliberate bet on a very large future.

The placement is Hong Kong’s largest-ever primary follow-on offering by a company listed there. Globally, it ranks third-largest this year, behind Alphabet’s $80 billion raise in June and Intel’s $15 billion sale in August. Sovereign wealth funds are reportedly among the early buyers. China International Capital Corp., HSBC, Morgan Stanley, and UBS are the bookrunners. Alibaba takes on a 90-day lockup.

Deep Dive: What Alibaba Actually Is in 2026

Alibaba started as an e-commerce business. That description no longer captures what it has become. The company reorganized itself into three core segments this year: Alibaba E-commerce Group (combining domestic commerce, international digital commerce, and Freshippo), AI Cloud and Compute Services (combining the cloud business with its T-Head semiconductor subsidiary), and AI Labs and Applications (housing the Qwen model family, QwenWork enterprise tools, and the consumer Qwen App).

The strategic intent is vertical integration across the full AI stack. T-Head designs proprietary AI chips. Alibaba Cloud runs the infrastructure those chips power. Qwen is the model layer trained on that infrastructure. QwenWork and the Qwen App are the enterprise and consumer application layers that monetize the models. More than 100,000 T-Head Zhenwu PPUs are now deployed on Alibaba Cloud’s public platform. Over 60% of T-Head’s compute capacity serves external customers, meaning the chip business is becoming a genuine revenue stream, not just an internal cost center.

The Qwen model family is the crown jewel of this architecture. Alibaba’s Tongyi Laboratory has released more than 400 open Qwen models since 2023, collectively surpassing one billion downloads and 200,000 derivative models on Hugging Face, overtaking Meta’s Llama as the default open foundation model globally. The strategy is deliberately two-tier: commoditize the middle of the market with open-weight models released under Apache 2.0 licensing, and monetize the top through proprietary, API-only frontier models on Alibaba Cloud. The latest Qwen3.8-Max has 2.4 trillion parameters. The consumer-facing Qwen App has reached 250 million users who have experienced AI-driven shopping features through integration with Taobao and Tmall.

Alibaba is also moving to monetize its open-source ecosystem directly. Reuters reported the company plans to ask major users of its Qwen open-source models to share a portion of the revenue they generate from products and services built on the technology. Previously, Alibaba charged only when models ran on its own cloud platform. That freemium-to-revenue-share shift is a meaningful change in the monetization calculus.

The Data Section

Let’s run the numbers that matter.

• Revenue (June quarter FY2027): 269 billion yuan, up 9% year over year. Full-year group revenue for FY2026 (ended March 2026) was approximately RMB 1.02 trillion ($148.4 billion).
• Cloud revenue (June quarter): External cloud revenue grew 45%, marking the fastest cloud growth in 22 quarters. AI-related product revenue posted triple-digit growth for the twelfth consecutive quarter.
• AI cloud EBITA margin: 11.6% in the June quarter, up from approximately 7% a year earlier, according to Nomura analysts. Nomura noted this growth was occurring alongside margin expansion, not at its expense.
• AI Labs and Applications segment: Revenue grew 16% to RMB 3.34 billion ($492 million), but adjusted EBITA losses widened to RMB 13.86 billion from RMB 3.22 billion a year earlier. A 330% increase in losses, driven by heavy model development spend and rising Qwen App inference costs.
• Capex: 68 billion yuan in the June quarter, up 75% year over year. Quarterly capex has reached nearly $10 billion. Alibaba has now spent roughly half its three-year RMB 380 billion ($53 billion) commitment.
• Free cash flow: Outflow of $6.6 billion in the June quarter.
• Net profit: 10.5 billion yuan ($1.6 billion), down 75%.
• AI payback timeline: CEO Eddie Wu indicated the company expects to break even on AI-related capex within three years at current average gross margins. The expected payback period for AI investments has shortened from 3 years to 2.5 years, citing surging demand.
• Model Studio (MaaS platform): Customer base expanded 8x year over year as of March 2026. AI-related cloud revenue was at a $5.2 billion annualized run rate as of March 2026.
• China AI cloud market share: Alibaba Cloud holds approximately 35.8% of China’s AI cloud market, per Omdia data cited by SCMP. ByteDance’s Volcano Engine sits in second place at 14.8%.
• Balance sheet: Market cap approximately $272 billion. Enterprise value approximately $254 billion. Debt-to-equity ratio of 0.24. Current ratio of 1.36.
• Asset pruning: CEO Eddie Wu has been selling non-core assets to help fund the AI pivot. The gaming arm Lingxi Games was sold to Trustar Capital in a deal reportedly worth at least $1.5 billion.

Is It Cheap?

This is where the Burry debate lands. And it depends almost entirely on which earnings number you use.

BABA’s trailing P/E ratio is distorted by the profit crater. The forward P/E, based on consensus estimates for the next 12 months, sits around 14.6x. Its PEG ratio is 0.44. Relative to its peer group average P/E of roughly 31x, BABA trades at less than half that multiple. Against the Global Multiline Retail industry average of approximately 19.8x, it is still a discount. Its 10-year historical average P/E is closer to 27x to 32x depending on the dataset. The stock is trading at roughly 46% below that long-run average.

Burry’s counterargument is that dilution changes the denominator. He framed share issuance as Alibaba’s “new paradigm,” which is the dilution bear case in plain English: if the company keeps printing equity to fund capex, per-share earnings growth gets harder to achieve even if the business grows. That concern is legitimate. The June quarter share count is already up 0.39% year over year, and the $10.2 billion placement will add roughly 710 million ordinary shares to a float of approximately 2.28 billion shares outstanding. That is dilution approaching 30% of current float if all shares are placed to new holders.

The bull response: Alibaba has historically been an aggressive buyback machine. If cloud revenue continues to accelerate and the AI payback timeline shortens as Wu projects, free cash flow turns positive again and dilution gets absorbed. Nomura’s post-earnings note argued cloud growth has not yet peaked, and the margin expansion occurring simultaneously with revenue growth is the most important signal in the data.

GuruFocus rates BABA as fairly valued at a GF Value of $117.78. Simply Wall St puts BABA’s P/E at 16.8x versus a fair P/E of 32x, implying meaningful upside if the earnings trajectory recovers. The consensus EPS growth estimate for the March 2027 fiscal year is 41.78%, followed by 26.64% in FY2028. If those estimates hold, the forward multiple compresses fast.

Bull / Base / Bear

Bull Case

Cloud revenue growing at 45% with margin expansion is not a story that stops quickly. Twelve consecutive quarters of triple-digit AI revenue growth tells you enterprise adoption is structural, not a blip. If the AI payback timeline compresses to 2.5 years as Wu guides, free cash flow recovers by late 2027 or early 2028. The Qwen open-source strategy mirrors what Red Hat did with Linux: give away the foundation, monetize the enterprise layer. A forward P/E of 14.6x on a business growing cloud at 45% is a number that does not survive contact with normalizing earnings. Add a potential regulatory re-rating as Beijing continues to signal support for private enterprise, and you get multiple expansion on top of earnings growth.

Base Case

Cloud growth slows to the 25% to 35% range as the initial AI adoption wave normalizes. The $10.2 billion placement funds the second half of the three-year capex plan. Earnings recover modestly in FY2028 as infrastructure spending plateaus. The stock re-rates from 14.6x forward to something closer to 18x to 20x as the market gets more confident in the FCF recovery. That implies a price target in the $130 to $150 range on current ADS levels, roughly 10% to 25% above where the stock traded before the placement announcement.

Bear Case

Burry’s framing is the bear case in concentrated form. If share issuance becomes habitual every time capex commitments outpace cash generation, per-share value erodes even as the enterprise grows. Add U.S.-China trade friction that constrains Alibaba’s international cloud expansion, a DeepSeek-style competitor that compresses model pricing faster than expected, and a domestic consumer that stays weak, and the earnings recovery pushed into FY2028 gets pushed again. The AI Labs segment is already losing RMB 13.86 billion per quarter and widening. If that does not convert to revenue, it is a structurally loss-making division grafted onto an otherwise profitable business.

Action Plan

This is not a stock you buy in a single tranche after an 8.57% down day driven by placement overhang. The overhang is real. 710 million new shares need to find permanent homes, and some of that institutional paper will be flipped. That pressure does not evaporate in a week.

For bargain hunters already holding BABA: the cloud data argues for patience, not selling. The dilution risk is real but manageable if the earnings recovery timeline holds. Do not add aggressively until the placement settles and you see how the stock finds support post-lockup.

For bargain hunters building a new position: consider a three-tranche scale-in. A first entry at current levels captures the post-placement dip. A second tranche if the stock pulls back further to the $95 to $100 range, which is where the forward P/E drops below 13x and starts screening as genuinely cheap even on conservative estimates. Hold the third tranche for the Q1 FY2028 earnings print, which should be the first quarter where the year-over-year capex comparison flattens and FCF inflects.

Position sizing matters. BABA carries China regulatory risk, U.S. ADR delisting risk in tail scenarios, and currency risk. Keep it a position, not a portfolio.

Cheap Investor Scorecard

• Cloud revenue growth rate: 45% in June quarter. Watch for acceleration or deceleration in the September quarter print. Target: sustained above 35%.
• AI-related product revenue: 12 consecutive quarters of triple-digit growth. The streak breaks when it breaks. Track quarterly.
• AI cloud EBITA margin: 11.6% and expanding. Watch for 15%+ as the signal that the infrastructure investment is paying back. Below 10% would be a warning.
• AI Labs losses: RMB 13.86 billion per quarter and widening 330% year over year. This needs to stabilize. Track the rate of change, not the absolute level.
• Free cash flow: Currently negative $6.6 billion per quarter. The inflection to positive FCF is the single most important trigger for multiple re-rating. First positive FCF quarter is the buy signal for the third tranche.
• AI payback timeline guidance: Management said 2.5 years from current investments. Any revision to 3+ years is a red flag.
• Share count: 2.28 billion outstanding pre-placement. Track dilution carefully. If issuance becomes quarterly, Burry’s framing wins.
• Qwen App monthly active users: 250 million in the June quarter. Watch for conversion to paid tiers or enterprise contracts. User growth without monetization is a cost, not an asset.
• China AI cloud market share: 35.8%. Watch for ByteDance and Huawei Cloud eroding that lead. Share below 30% changes the competitive thesis.
• Capex plan revision: The original commitment was $53 billion over three years. Reports suggest Alibaba is considering raising it to $69 billion. If a new, larger commitment is announced, recalibrate the FCF timeline accordingly.

Bottom Line

If Alibaba’s cloud growth holds above 35%, the AI payback timeline compresses to 2.5 years as guided, and share issuance stays a one-time event rather than a recurring habit, BABA at a forward P/E below 15x is genuinely cheap for a business with this growth profile. That is the conditional case for owning it.

If Burry is right that equity issuance has become structural, and if the AI Labs division continues to widen losses without generating commercial revenue at scale, then the earnings recovery story gets deferred indefinitely and the discount is a value trap, not a bargain.

The market is pricing in significant uncertainty. That is usually where interesting situations live. Watch the September quarter cloud number. Watch the FCF line. Watch whether a second share placement appears in the next 18 months. The answers to those three questions will tell you which version of Alibaba you actually own.

The Cheap Investor