Baidu, Inc.
BIDU has real AI infrastructure momentum, but the core ads business is still shrinking and the stock needs proof that AI growth can offset that erosion before a more aggressive stance is justified.
The market is not wrong to discount Baidu. The legacy business is still shrinking, and management is telling investors not to expect near-term relief in advertising. FY2025 revenue was $19.2B, down 3.0% YoY. Q2 2026 revenue was RMB 31.3B, down 4% YoY. On the call, Rong Luo said, "Given these dynamics are likely to persist, we expect our advertising business to remain under pressure in the second half." That sentence matters more than the post-dip headline noise. A company with a pressured core should not get a full AI premium just because one segment is growing quickly.
The reason not to be bearish here is that the AI business is no longer trivial. AI Cloud Infra grew 50% YoY, GPU Cloud grew 283%, and Qianfan token revenue >9x YoY from external customers. That is real adoption, and it suggests Baidu has moved beyond a pure search story into a domestic AI infrastructure platform with improving monetization. AI-powered business at RMB 12.5B, roughly half of Baidu General Business revenue, also shows the mix shift is already underway. If that growth sustains into 2027, the current 1.85x P/S could look too low.
The problem is that investors still need proof of earnings quality and cash generation. FY2025 operating margin fell to 15.1% from 16.0%. Q2 EBITDA margin was 33.5%, down 270 bps sequentially from 36.2%. Free cash flow is weak, with the snapshot showing negative $0.5B FCF, and deep-dive data showing only $0.5B FCF versus $0.9B capex. The absurd trailing 796.11x P/E is not useful, but it does capture one truth: current earnings are not a clean base for valuation. The stock should be valued on forward normalization, and that normalization still depends on two things the market cannot yet underwrite with confidence, ad stabilization and durable AI margin contribution.
At $90.87, the setup is no longer expensive. The shares are 45.0% below the 52-week high and 27.1% below the 200-day average. That makes downside less open-ended than it was near the highs, but it does not create a buy by itself. This case depends on three specific milestones: AI Cloud maintaining at least current growth, ad declines easing from current pressure, and Apollo Go showing breakeven progress in more cities during 2H26. Until at least one of those translates into cleaner consolidated revenue and cash flow, HOLD is the disciplined call.
AI Cloud acceleration in 2H26, ad stabilization, and Apollo Go breakeven progress drive a sharper rerating from a stock already 45% below its high.
AI Cloud remains strong, ad declines persist but do not worsen, and the stock recovers modestly toward the quant target from an oversold setup.
Ads remain under pressure into 2027, AI Cloud growth slows from current +50%, and weak cash generation prevents a durable re-rating.
Quant call (HOLD) diverges from analyst consensus (Strong Buy: 3 buy / 2 hold / 0 sell) — resolve explicitly.
You have power over your mind — not outside events. Realize this, and you will find strength.
Forward earnings power depends less on the meaningless trailing 796.11x P/E and more on whether the company can rebuild normalized profit through AI infrastructure while stabilizing search. If Baidu can sustain 50%+ AI Cloud Infra growth and keep GPU Cloud triple digit growth while ad declines moderate, the market can look through the current depressed revenue base and support something closer to the $102.88 base value; if ads keep shrinking into 2027, even 1.85x sales may not be cheap enough. The moat is shifting from legacy search distribution toward AI infrastructure, foundation models, and local-language enterprise tooling, but that moat looks narrower today than it did three years ago because AI assistants and rival ecosystems can disintermediate search. Regulation cuts both ways: Apollo Go expansion and China tech policy can help, but the recent operating adjustments in domestic cities show autonomy and platform businesses still carry policy friction.
- Core fundamental split is widening: FY2025 revenue was $19.2B, down 3.0% YoY, while Q2 2026 total revenue was RMB 31.3B, down 4% YoY because online marketing stayed weak.
- AI engine is real: AI Cloud Infra +50% YoY, GPU Cloud +283% YoY, and AI-powered business reached RMB 12.5B, about half of Baidu General Business revenue.
- Setup is washed out but not yet confirmed: shares sit at $90.87, 27.1% below the 200-day average and 45.0% below the 52-week high, with the quant base target at $102.88, only 13.2% upside.
- AI Cloud Infra +50% YoY in Q2 2026 shows Baidu has a real demand signal in enterprise AI, not just a concept trade.
- GPU Cloud +283% YoY improved mix and profit, suggesting the AI build is shifting toward monetizable infrastructure rather than purely experimental spending.
- Qianfan token revenue >9x YoY from external customers indicates third-party adoption of Baidu's AI stack, which matters more than internal usage.
- Apollo Go ~1 million rides in Q2 and over 23 million cumulative rides by June show the autonomous platform still has operating scale despite temporary regulatory disruption.
- The legacy engine is still deteriorating: FY2025 revenue declined 3.0%, and management said, "Given these dynamics are likely to persist, we expect our advertising business to remain under pressure in the second half."
- Profit quality is mixed: FY2025 operating margin fell to 15.1% from 16.0%, and Q2 EBITDA margin dropped to 33.5%, down 270 bps versus Q1's 36.2%.
- Cash conversion is weak for a company trying to claim AI leadership, with FCF negative $0.5B in the financial snapshot and deep-dive cash flow showing only $0.5B FCF against $0.9B capex.
- The stock remains technically fragile after a -28.89% 3-month move, and options positioning is in a negative gamma regime with zero-gamma at $99.5, above spot.
| Metric | Value | Context |
|---|---|---|
| Current price | $90.87 | Reference price for the engine target and scenario framing |
| Price target | $102.88 | Quant engine base target, 13.2% upside, Medium confidence |
| FY2025 revenue | $19.2B | -3.0% YoY, shows consolidated contraction |
| FY2025 EPS | $1.97 | +50.5% YoY, but distorted relative to the trailing 796.11x P/E |
| FY2025 operating margin | 15.1% | Down from 16.0% in the prior fiscal year |
| Valuation | P/E 796.11, P/S 1.85, P/B 6.12 | P/S sits below peer average 2.7, while P/E is not decision-useful |
| Balance sheet | Debt/Equity 0.35, Debt/EBITDA 0.57, Current ratio 1.85 | Leverage is manageable despite weaker cash generation |
| Free cash flow | $-0.5B | Weak cash conversion in the FY2025 snapshot |
| Q2 2026 revenue and EBITDA | RMB 31.3B revenue, RMB 10.97B EBITDA | Revenue -4% YoY, EBITDA margin 33.5% |
| AI growth indicators | AI Cloud Infra +50% YoY, GPU Cloud +283% YoY | Strongest part of the business and the main source of forward upside |
| 52-week position | $84.82 to $165.30 | Current price is -45.0% vs the 52-week high |
The biggest risk is that the ads business, still central to consolidated earnings, remains under pressure beyond 2H26, leaving AI growth unable to offset a company with FY2025 revenue already down 3.0% and Q2 2026 revenue down 4% YoY.
The key catalyst is the next earnings update and whether AI Cloud growth accelerates enough to offset further ad weakness, especially after management said AI Cloud Infra has "the potential for further acceleration" in 2H26.
Ownership is not crowded in the classic U.S. mega-cap sense, with the top three institutions holding only 14.8%, while reported short interest is modest at 3.11% of shares and 3.8 days-to-cover after a -15.5% change in short shares. That means there is no obvious squeeze setup. Options positioning is less helpful near term, with negative gamma and zero-gamma at $99.5, which can amplify volatility until the stock reclaims that level.
Now large enough to matter, supported by AI Cloud Infra growth and token monetization.
Still the main drag, with management guiding the business to remain under pressure through 2H26.
Q2 2026 grew +50% YoY, with management guiding strong growth and possible acceleration in 2H26.
Q2 volume was about 1 million rides and management expects more cities to reach breakeven in 2H26.
AI Cloud kept growing fast, but ads stay under pressure in 2H.
Management deserves partial credit, not a full pass. On the 2024 Q3 call, they warned that search monetization would not be heavily pushed near term and that margins would be in adjustment, and that caution proved directionally right: FY2025 revenue fell 3.0%, FY2025 operating margin slipped to 15.1% from 16.0%, and Q2 2026 revenue was still down 4% YoY. They also said in 2024, "We believe that following this period of adjustment, we expect to start to seeing improvements in our advertising business next year." That did not materialize cleanly, since 2026 guidance now says ads remain pressured in 2H26. The better execution came on AI: AI Cloud was +11% YoY in 2024 Q3 and is now +50% YoY in AI Cloud Infra, GPU Cloud is +283%, and Qianfan token revenue is >9x YoY. The beat-miss record is solid but not spotless, with 6 beats and 2 misses across the last eight reported quarters, including the latest -6.4% EPS miss on 2026-05-18. Net result: credible on identifying ad pressure early and building AI products, less credible on the timing of ad recovery.
Based on current demand trends, our customer pipelines and these differentiated advantages, so we feel confident that AI Cloud Infra can maintain strong growth in the second half with the potential for further acceleration.
Given these dynamics are likely to persist, we expect our advertising business to remain under pressure in the second half.
We are confident that ride volume will regain momentum over the coming quarters as we steadily ramp up operations and pursue further expansion.
Total revenue of Baidu was RMB 31.3 billion, decreasing 2% quarter-over-quarter and 4% year-over-year.
We currently expect the conversion to become effective within this year, subject to the approval of the shareholders and the Hong Kong Stock Exchange.
Baidu's moat still rests on Chinese-language search data, distribution, and a full-stack domestic AI stack that includes models, cloud infrastructure, and Kunlunxin chips. That matters because compute supply is constrained and management says Kunlunxin demand remains robust, which can support pricing and customer stickiness. The durability is mixed: cloud and enterprise AI can deepen the moat over the next three to five years, but the legacy search moat is weaker as user behavior shifts toward AI-native interfaces and content ecosystems outside classic search.
Capital allocation is acceptable but not yet shareholder-friendly enough to offset execution risk. Deep-dive data shows $0.5B FCF, $0.9B capex, $0.0B buybacks, and $0.0B dividends, so current cash deployment is going into AI buildout rather than direct returns. That is rational if AI Cloud keeps compounding at current rates, but compared with management's 2024 commentary around roughly US$1B annual repurchases on average, the current lack of buyback support reduces downside cushioning.
Rated HOLD, Medium conviction. Baidu has a legitimate AI business, and the stock has already fallen far enough that valuation on sales is no longer demanding. But the company is still trying to replace a weakening ad engine with a newer AI engine that is growing fast yet has not fully proven consolidated earnings power or cash conversion. A move to BUY would require evidence that ad pressure is bottoming and AI growth is starting to lift total revenue and free cash flow together. A move to SELL would require signs that AI Cloud growth is slowing while ad declines persist into 2027.
≣ Full research note tap to expand
Executive Summary
HOLD, Medium conviction. Rated HOLD, Medium conviction. The stock is down 45.0% from its 52-week high and now screens cheap on sales at 1.85x P/S, but that discount exists for a reason: FY2025 revenue fell 3.0%, Q2 2026 revenue fell 4% YoY, and management explicitly said ads will remain pressured through 2H26. AI Cloud is growing fast enough to matter, with AI Cloud Infra +50% YoY and GPU Cloud +283%, but not yet fast enough to cleanly re-rate the whole company while free cash flow remains weak.
Investment Thesis
The market is not wrong to discount Baidu. The legacy business is still shrinking, and management is telling investors not to expect near-term relief in advertising. FY2025 revenue was $19.2B, down 3.0% YoY. Q2 2026 revenue was RMB 31.3B, down 4% YoY. On the call, Rong Luo said, "Given these dynamics are likely to persist, we expect our advertising business to remain under pressure in the second half." That sentence matters more than the post-dip headline noise. A company with a pressured core should not get a full AI premium just because one segment is growing quickly.
The reason not to be bearish here is that the AI business is no longer trivial. AI Cloud Infra grew 50% YoY, GPU Cloud grew 283%, and Qianfan token revenue >9x YoY from external customers. That is real adoption, and it suggests Baidu has moved beyond a pure search story into a domestic AI infrastructure platform with improving monetization. AI-powered business at RMB 12.5B, roughly half of Baidu General Business revenue, also shows the mix shift is already underway. If that growth sustains into 2027, the current 1.85x P/S could look too low.
The problem is that investors still need proof of earnings quality and cash generation. FY2025 operating margin fell to 15.1% from 16.0%. Q2 EBITDA margin was 33.5%, down 270 bps sequentially from 36.2%. Free cash flow is weak, with the snapshot showing negative $0.5B FCF, and deep-dive data showing only $0.5B FCF versus $0.9B capex. The absurd trailing 796.11x P/E is not useful, but it does capture one truth: current earnings are not a clean base for valuation. The stock should be valued on forward normalization, and that normalization still depends on two things the market cannot yet underwrite with confidence, ad stabilization and durable AI margin contribution.
At $90.87, the setup is no longer expensive. The shares are 45.0% below the 52-week high and 27.1% below the 200-day average. That makes downside less open-ended than it was near the highs, but it does not create a buy by itself. This case depends on three specific milestones: AI Cloud maintaining at least current growth, ad declines easing from current pressure, and Apollo Go showing breakeven progress in more cities during 2H26. Until at least one of those translates into cleaner consolidated revenue and cash flow, HOLD is the disciplined call.
Key Metrics
| Metric | Value | Context |
|---|---|---|
| Current price | $90.87 | Reference price for the engine target and scenario framing |
| Price target | $102.88 | Quant engine base target, 13.2% upside, Medium confidence |
| FY2025 revenue | $19.2B | -3.0% YoY, shows consolidated contraction |
| FY2025 EPS | $1.97 | +50.5% YoY, but distorted relative to the trailing 796.11x P/E |
| FY2025 operating margin | 15.1% | Down from 16.0% in the prior fiscal year |
| Valuation | P/E 796.11, P/S 1.85, P/B 6.12 | P/S sits below peer average 2.7, while P/E is not decision-useful |
| Balance sheet | Debt/Equity 0.35, Debt/EBITDA 0.57, Current ratio 1.85 | Leverage is manageable despite weaker cash generation |
| Free cash flow | $-0.5B | Weak cash conversion in the FY2025 snapshot |
| Q2 2026 revenue and EBITDA | RMB 31.3B revenue, RMB 10.97B EBITDA | Revenue -4% YoY, EBITDA margin 33.5% |
| AI growth indicators | AI Cloud Infra +50% YoY, GPU Cloud +283% YoY | Strongest part of the business and the main source of forward upside |
| 52-week position | $84.82 to $165.30 | Current price is -45.0% vs the 52-week high |
Financial Strength
The balance sheet is not the problem. Debt/Equity at 0.35, Debt/EBITDA at 0.57, and a 1.85 current ratio give Baidu room to keep investing through the transition. The issue is cash conversion and margin direction. Operating margin slipped in FY2025, Q2 EBITDA margin compressed sequentially, and free cash flow is weak relative to capex. That combination argues the company can fund itself, but not yet that it is translating AI momentum into consistently higher-quality earnings.
Competitive Position
Baidu's moat still rests on Chinese-language search data, distribution, and a full-stack domestic AI stack that includes models, cloud infrastructure, and Kunlunxin chips. That matters because compute supply is constrained and management says Kunlunxin demand remains robust, which can support pricing and customer stickiness. The durability is mixed: cloud and enterprise AI can deepen the moat over the next three to five years, but the legacy search moat is weaker as user behavior shifts toward AI-native interfaces and content ecosystems outside classic search.
Management & Guidance
Current guidance is mixed and should be taken literally. Management is clearly bullish on AI Cloud, with Dou Shen saying, "Based on current demand trends, our customer pipelines and these differentiated advantages, so we feel confident that AI Cloud Infra can maintain strong growth in the second half with the potential for further acceleration." At the same time, management was equally clear on the core drag: ads will remain pressured in 2H26. Apollo Go is expected to regain momentum, and more cities are expected to approach breakeven in 2H26, but recent regulatory-related operating adjustments are a reminder that the timeline is not fully in management's control. Given the recent earnings record, including the latest -6.4% EPS miss, management deserves moderate credibility, not full benefit of the doubt.
Segment Analysis
- AI-powered business (RMB 12.5B, about 50% of Baidu General Business revenue): Now large enough to matter, supported by AI Cloud Infra growth and token monetization.
- Online marketing / advertising (Not separately disclosed in the provided data): Still the main drag, with management guiding the business to remain under pressure through 2H26.
- AI Cloud Infrastructure (Not separately disclosed in the provided data): Q2 2026 grew +50% YoY, with management guiding strong growth and possible acceleration in 2H26.
- Apollo Go / autonomous mobility (Not separately disclosed in the provided data): Q2 volume was about 1 million rides and management expects more cities to reach breakeven in 2H26.
Capital Allocation
Capital allocation is acceptable but not yet shareholder-friendly enough to offset execution risk. Deep-dive data shows $0.5B FCF, $0.9B capex, $0.0B buybacks, and $0.0B dividends, so current cash deployment is going into AI buildout rather than direct returns. That is rational if AI Cloud keeps compounding at current rates, but compared with management's 2024 commentary around roughly US$1B annual repurchases on average, the current lack of buyback support reduces downside cushioning.
Management Execution & Track Record
Management deserves partial credit, not a full pass. On the 2024 Q3 call, they warned that search monetization would not be heavily pushed near term and that margins would be in adjustment, and that caution proved directionally right: FY2025 revenue fell 3.0%, FY2025 operating margin slipped to 15.1% from 16.0%, and Q2 2026 revenue was still down 4% YoY. They also said in 2024, "We believe that following this period of adjustment, we expect to start to seeing improvements in our advertising business next year." That did not materialize cleanly, since 2026 guidance now says ads remain pressured in 2H26. The better execution came on AI: AI Cloud was +11% YoY in 2024 Q3 and is now +50% YoY in AI Cloud Infra, GPU Cloud is +283%, and Qianfan token revenue is >9x YoY. The beat-miss record is solid but not spotless, with 6 beats and 2 misses across the last eight reported quarters, including the latest -6.4% EPS miss on 2026-05-18. Net result: credible on identifying ad pressure early and building AI products, less credible on the timing of ad recovery.
Positioning & Flows
Ownership is not crowded in the classic U.S. mega-cap sense, with the top three institutions holding only 14.8%, while reported short interest is modest at 3.11% of shares and 3.8 days-to-cover after a -15.5% change in short shares. That means there is no obvious squeeze setup. Options positioning is less helpful near term, with negative gamma and zero-gamma at $99.5, which can amplify volatility until the stock reclaims that level.
Bull Case
- AI Cloud Infra +50% YoY in Q2 2026 shows Baidu has a real demand signal in enterprise AI, not just a concept trade.
- GPU Cloud +283% YoY improved mix and profit, suggesting the AI build is shifting toward monetizable infrastructure rather than purely experimental spending.
- Qianfan token revenue >9x YoY from external customers indicates third-party adoption of Baidu's AI stack, which matters more than internal usage.
- Apollo Go ~1 million rides in Q2 and over 23 million cumulative rides by June show the autonomous platform still has operating scale despite temporary regulatory disruption.
Bear Case
- The legacy engine is still deteriorating: FY2025 revenue declined 3.0%, and management said, "Given these dynamics are likely to persist, we expect our advertising business to remain under pressure in the second half."
- Profit quality is mixed: FY2025 operating margin fell to 15.1% from 16.0%, and Q2 EBITDA margin dropped to 33.5%, down 270 bps versus Q1's 36.2%.
- Cash conversion is weak for a company trying to claim AI leadership, with FCF negative $0.5B in the financial snapshot and deep-dive cash flow showing only $0.5B FCF against $0.9B capex.
- The stock remains technically fragile after a -28.89% 3-month move, and options positioning is in a negative gamma regime with zero-gamma at $99.5, above spot.
Valuation & Price Target
Engine price target $102.88 (+13.2% vs current), Medium confidence.
- P/E Multiple: $14.38 (1% weight)
- P/S Multiple: $111 (20% weight)
- Quality-adjusted: $77.86 (20% weight)
- DCF: $358.85 (2% weight)
The rating stays aligned with the quant baseline at HOLD, and the reason is straightforward. The engine's price target is $102.88, only 13.2% above the current $90.87, which is not enough upside to override a business with FY2025 revenue down 3.0%, FY2025 operating margin down to 15.1%, and a latest-quarter revenue decline of 4% YoY. The valuation blend itself argues for caution: DCF is $358.85, but the market-based methods are far lower at $14.38 on P/E, $111 on P/S, and $77.86 quality-adjusted, while peer P/E dispersion is a very wide 17.9x across 6 peers. That spread says the stock is highly sensitive to method choice because earnings are currently distorted and comparables are noisy. Sell-side consensus is Strong Buy from only 5 firms, which is low-signal context and not persuasive against explicit company guidance that ads remain pressured in 2H26.
Risk Assessment
Risk score 49/100 (Moderate). The biggest risk is that the ads business, still central to consolidated earnings, remains under pressure beyond 2H26, leaving AI growth unable to offset a company with FY2025 revenue already down 3.0% and Q2 2026 revenue down 4% YoY.
The Bottom Line
Rated HOLD, Medium conviction. Baidu has a legitimate AI business, and the stock has already fallen far enough that valuation on sales is no longer demanding. But the company is still trying to replace a weakening ad engine with a newer AI engine that is growing fast yet has not fully proven consolidated earnings power or cash conversion. A move to BUY would require evidence that ad pressure is bottoming and AI growth is starting to lift total revenue and free cash flow together. A move to SELL would require signs that AI Cloud growth is slowing while ad declines persist into 2027.
Go deeper in the app
Live charts, options analytics, the full earnings-call breakdown, price alerts, and every other stock — updated in real time.
Get Tickr