Baidu
Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat Score
Chinese search incumbent transitioning to AI cloud + autonomous (Apollo) platform — moats real but eroding from generative search disruption and Bytedance / Tencent competition. Q2 2026 kept Core AI at 50% of General Business (from 52% in Q1) even as company revenue fell 4%; GPU cloud accelerated to +283%, while Online Marketing −19% confirmed the search franchise is still shrinking.
Baidu's moat is Chinese search history + AI cloud (ERNIE) + Apollo autonomous franchise — search eroding fast, AI/autonomy emerging, valuation prices in failure:
- ERNIE LLM and AI Cloud: ERNIE 5.1 (May 2026) ranks first among Chinese models on LMArena text and fourth globally on search. Q2 AI Cloud Infra revenue was RMB7.3B (+50% YoY) with GPU cloud +283% (accelerating from +184% in Q1); Core AI-powered business RMB12.5B, 50% of General Business. The mix shift is real; the company print is still down because ads fell faster.
- Apollo Robotaxi Platform: Apollo Go now spans 28 cities with 350M+ autonomous km (240M+ fully driverless). Q2 added fully driverless commercial operations in Dubai (including via Uber), open-road testing in London with Uber/Lyft, Hong Kong's first fully driverless test permits, and Switzerland testing with PostBus. Still loss-making; commercial viability depends on unit economics and overseas regulatory pace.
- Search Heritage and Distribution: Baidu remains the dominant Chinese search engine with ~70% domestic share, but Online Marketing Services fell 19% YoY in Q2 2026 to RMB13.1B — slightly less severe than Q1's −22%, not a reversal. Users continue migrating to Bytedance / Xiaohongshu / WeChat; the decline is structural.
Moat Verdict
Baidu's search moat is the AI-vulnerable canonical example — generative AI directly disrupts the search-advertising franchise, and Q2's −19% Online Marketing print (on a −4% company) confirmed the pace. The thesis question is whether ERNIE + Apollo offsets fast enough; Core AI holding ~50% of General Business is progress, GPU cloud +283% is the tell, but valuation still prices in failure with optionality on transition success.
46.5 resilient · 51.5 vulnerable · 80/20 = 47.5 · = 48
Open a moat to read its note.
Search interface learning is real but generative AI and short-video are eroding the search habit.
ERNIE 5.1 + full-stack AI cloud platform is real differentiated business logic for B2B inference; LMArena rankings support model quality claims.
Decades of Chinese web crawl data is genuinely unique and feeds ERNIE training; declining utility as search habit erodes.
Chinese AI engineering bench is real and durable; ERNIE team rivals Alibaba Qwen and Tencent Hunyuan.
Search + AI cloud + Apollo + iQiyi creates breadth but not deep bundling vs Tencent / Alibaba ecosystems.
Chinese search query + crawl + maps data is unique but commoditising as generative search reshapes user behaviour.
Chinese internet content licences and Apollo robotaxi permits create real new-entrant friction; overseas permits (UK, UAE, Switzerland, Hong Kong) are incremental but early.
Search network effects (more queries → better ranking) eroding on user migration; Apollo network effects emerging across 28 cities / 350M km but still small.
Search advertising integrations real but commoditising (−19% YoY); AI cloud embedment building on enterprise GPU demand.
Search no longer the system of record for many Chinese users; AI cloud emerging in B2B but small base.
Scale here is the two-sided network itself, already rated under networkEffects; crediting it again as a cost lead would double-count the same evidence.
The consumer name is a product of the network, not an independent source of pricing power; rated under networkEffects rather than here.
Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat Score
Chinese search incumbent transitioning to AI cloud + autonomous (Apollo) platform — moats real but eroding from generative search disruption and Bytedance / Tencent competition. Q2 2026 kept Core AI at 50% of General Business (from 52% in Q1) even as company revenue fell 4%; GPU cloud accelerated to +283%, while Online Marketing −19% confirmed the search franchise is still shrinking.
Growth Score
Q2 2026 company revenue RMB31.3B (−4% YoY), missing consensus, as Online Marketing −19% outran Core AI +25%. General Business −4% after Q1's +2%. GPU cloud +283% and Apollo's 28-city footprint are the offset; they are not yet large enough to grow the whole company. Multi-year CAGR hinges on AI cloud sustaining 30%+ while search decline moderates from the high-teens.
Valuation Score
At ~$93 (Aug 19, 2026) after an ~11% drop on the Q2 miss, Baidu trades well below the $130 base and ~33% above the $70 bear — closer to the cheap end of the corridor than at the August 10 ~$109 print. Cash and investments of $42B exceed the ~$32B market cap (gross of debt and including long-term investments). Operating business still screens at a mid-teens P/E on depressed earnings; the Aug 18 print did not de-risk the transition, it cheapened the option.
The Transition-to-AI Moat
Baidu's moat is Chinese search history + AI cloud (ERNIE) + Apollo autonomous franchise — search eroding fast, AI/autonomy emerging, valuation prices in failure:
- ERNIE LLM and AI Cloud: ERNIE 5.1 (May 2026) ranks first among Chinese models on LMArena text and fourth globally on search. Q2 AI Cloud Infra revenue was RMB7.3B (+50% YoY) with GPU cloud +283% (accelerating from +184% in Q1); Core AI-powered business RMB12.5B, 50% of General Business. The mix shift is real; the company print is still down because ads fell faster.
- Apollo Robotaxi Platform: Apollo Go now spans 28 cities with 350M+ autonomous km (240M+ fully driverless). Q2 added fully driverless commercial operations in Dubai (including via Uber), open-road testing in London with Uber/Lyft, Hong Kong's first fully driverless test permits, and Switzerland testing with PostBus. Still loss-making; commercial viability depends on unit economics and overseas regulatory pace.
- Search Heritage and Distribution: Baidu remains the dominant Chinese search engine with ~70% domestic share, but Online Marketing Services fell 19% YoY in Q2 2026 to RMB13.1B — slightly less severe than Q1's −22%, not a reversal. Users continue migrating to Bytedance / Xiaohongshu / WeChat; the decline is structural.
Moat Verdict
Baidu's search moat is the AI-vulnerable canonical example — generative AI directly disrupts the search-advertising franchise, and Q2's −19% Online Marketing print (on a −4% company) confirmed the pace. The thesis question is whether ERNIE + Apollo offsets fast enough; Core AI holding ~50% of General Business is progress, GPU cloud +283% is the tell, but valuation still prices in failure with optionality on transition success.
46.5 resilient · 51.5 vulnerable · 80/20 = 47.5 · = 48
Open a moat to read its note.
Search interface learning is real but generative AI and short-video are eroding the search habit.
ERNIE 5.1 + full-stack AI cloud platform is real differentiated business logic for B2B inference; LMArena rankings support model quality claims.
Decades of Chinese web crawl data is genuinely unique and feeds ERNIE training; declining utility as search habit erodes.
Chinese AI engineering bench is real and durable; ERNIE team rivals Alibaba Qwen and Tencent Hunyuan.
Search + AI cloud + Apollo + iQiyi creates breadth but not deep bundling vs Tencent / Alibaba ecosystems.
Chinese search query + crawl + maps data is unique but commoditising as generative search reshapes user behaviour.
Chinese internet content licences and Apollo robotaxi permits create real new-entrant friction; overseas permits (UK, UAE, Switzerland, Hong Kong) are incremental but early.
Search network effects (more queries → better ranking) eroding on user migration; Apollo network effects emerging across 28 cities / 350M km but still small.
Search advertising integrations real but commoditising (−19% YoY); AI cloud embedment building on enterprise GPU demand.
Search no longer the system of record for many Chinese users; AI cloud emerging in B2B but small base.
Scale here is the two-sided network itself, already rated under networkEffects; crediting it again as a cost lead would double-count the same evidence.
The consumer name is a product of the network, not an independent source of pricing power; rated under networkEffects rather than here.
Growth Analysis
Growth Drivers
Key Risk
If AI cloud growth fails to keep offsetting the high-teens search decline through 2027 and Apollo commercial deployment is delayed by Chinese or overseas regulatory pace, Baidu's revenue keeps compressing and the transition thesis collapses, leaving the equity as a melting search asset.
Score Derivation
57.5 base + 1.3 trajectory − 4 margin − 5 risk = 50
Base 57.5 (1-5% CAGR midpoint 3%) + 1.3 trajectory (AI cloud and Apollo accelerating vs search decelerating) − 4 margin compression on AI investment and ad mix − 5 moderate residual transition/geopolitical risk = 50. The Q2 company decline is charged in the lower CAGR, not double-counted in severity.
Price Scenarios (12–24 Months)
Valuation Multiples
| Forward P/E (FY26) | ~13× |
| Cash & investments / mkt cap | >1× |
| Price / Sales (FY26) | ~1.7× |
| Non-GAAP diluted EPS (Q2) | $1.06 |
| OCF | RMB3.4B |
Drawdown from ~$109 → ~$93 restored a deeper cash-heavy discount. The Q2 miss (revenue −4%, ads −19%) is why; GPU cloud +283% is why the option is still live. Next leg still requires AI cloud + Apollo to outgrow the search hole.
Approximate figures as of August 19, 2026.
Where We Are vs Targets
Loading live price…
Search decline stays double-digit, AI cloud growth disappoints, Apollo commercial viability delayed, multiple stays at ~8× depressed earnings.
- Online Marketing declines remain ~20%+ as users shift to short-video and Xiaohongshu
- AI cloud growth slows below 25% on competitive pressure from Alicloud + Huawei
- Apollo robotaxi commercial deployment delayed beyond 2027; London/Dubai stall on regulation
AI cloud + autonomy offset search decline, total revenue grows low- to mid-single-digits, FY28 EPS reaches ~$11-12, multiple expands to ~11×.
- AI cloud reaches ~$9B+ run-rate by FY28 with 25%+ growth
- Apollo robotaxi achieves break-even in select China cities by FY28; overseas still pre-revenue
- Search advertising decline moderates from the high-teens toward high-single-digits as migration normalises
ERNIE establishes dominant Chinese LLM franchise, Apollo scales internationally on RT6 economics, FY29 EPS reaches $15+, multiple rerates to 12× on franchise quality reassessment.
- ERNIE sustains top-tier Chinese LLM share; AI cloud mix exceeds 60% of General Business
- Apollo robotaxi reaches commercial profitability in China and launches public rides in London/Dubai
- Capital return reignites with buybacks materially exceeding the $259M YTD 2026 pace