Baidu

The pivot crossed its threshold: AI-powered revenue is now the majority of Baidu's core business at RMB 13.6B (+49% YoY) with AI Cloud Infra +79% and GPU Cloud +184%, enough to drag the core segment back to positive growth. Apollo Go is scaling into a genuine second business - 3.2M fully driverless rides in the quarter, 27 cities, over 220M fully driverless kilometres - and the whole transition is funded from a ~US$27.9B net cash position while the equity trades at roughly 2x sales.

• AI-powered business hit 52% of Baidu General Business revenue for the first time (RMB 13.6B, +49% YoY), so the growth engine is now larger than the declining one inside the core segment
• AI Cloud Infra grew 79% YoY to RMB 8.8B with GPU Cloud up 184% - demand-side evidence that Baidu's full-stack position (Ernie models plus in-house Kunlunxin accelerators) is winning enterprise workloads in a supply-constrained Chinese compute market
• Baidu General Business returned to positive revenue growth (+2% YoY) and held a 13% GAAP operating margin while absorbing the AI investment cycle
• Apollo Go is operating at real scale with a widening moat - 3.2M fully driverless rides in Q1, peak >350k weekly, 27 cities, >220M fully driverless kilometres, and international expansion into Dubai, Switzerland and planned London testing with Uber and Lyft
• Balance sheet gives it time: total cash and investments RMB 279.3B (US$40.49B) against RMB 86.8B of borrowings, so ~US$27.9B net, versus a US$38.35B market cap - and management is buying back stock (US$172M in Q1)
• The pending dual-primary Hong Kong listing plus a 10% repurchase mandate broadens the shareholder base and adds a mechanical support under the shares

Consolidated revenue is still shrinking (-1.2% YoY in Q1 2026, -3.0% in FY2025) and the AI pivot is being paid for out of margin: gross margin fell ~7.2pts to 38.9%, operating margin from 14% to 10%, net income attributable dropped 55% YoY, capex tripled sequentially and free cash flow went to -RMB 3.2B. FY2025 net income had already collapsed 76% to RMB 5.6B. Shareholders now also face a proposed mandate to issue up to 20% more shares.

• The mix shift is not yet a growth story - total revenue fell 1.2% YoY because legacy business dropped 29% and iQIYI 13%, so the +49% AI growth is largely substituting for revenue being lost, not adding to it
• Margin structure is deteriorating fast: gross margin 38.9% versus 46.1% a year ago and operating margin 10% versus 14%, driven by AI Cloud cost of revenue - GPU-heavy cloud is a structurally lower-margin business than search advertising
• Earnings quality and level both fell - net income attributable to Baidu was RMB 3,445M, down 55% YoY, and the year-ago comparison was itself inflated by RMB 4,487M of other income that shrank to RMB 626M; FY2025 net income of RMB 5,589M was already down 76% from RMB 23,760M in FY2024
• Cash generation has inverted: capex of RMB 5,916M was more than double operating cash flow of RMB 2,670M, producing -RMB 3,246M of free cash flow, and the AI Cloud and robotaxi buildouts have no disclosed spending ceiling
• Dilution is on the ballot - the 2026-08-26 EGM votes on an issuance mandate for up to 20% of issued shares plus a new 2026 Share Incentive Plan, against a 10% repurchase mandate
• iQIYI is now a drag rather than a diversifier, swinging to a RMB 228M operating loss on revenue down 13% YoY, while its debt sits inside the consolidated balance sheet
• The apparent cheapness rests on a broad cash definition - 'total cash and investments' of RMB 279.3B includes long-term time deposits, held-to-maturity instruments and adjusted long-term investments, so the liquid net cash backing the valuation is smaller than the headline
• No numeric guidance and no visibility into Q2 until 2026-08-18, in a business where the two swing factors (ad market, cloud pricing) are both outside management's control