Inve US ▾
BIBIDU

Baidu (BIDU) Q1 FY26 earnings call: guidance and Q&A

Earnings call· Mar 2026 (Q1 FY26)· Information Technology· May 18, 2026

Source: earnings call transcript

Executive summary

Baidu Q1 FY26 — AI-powered business crosses half of revenue as GPU cloud accelerates

Baidu's quarter marks an inflection: the AI-powered mix now outweighs the legacy marketing engine, and management is leaning into a durable margin-mix shift as higher-margin GPU cloud scales against still-declining online marketing. The forward stance is reinvestment-heavy — funding frontier models, full-stack silicon, and global robotaxi expansion — while framing operating cash flow and cost discipline as proof the pivot is financially healthy.

Highlights

5
  • Baidu Core AI-powered business revenue exceeded RMB 13 billion, up 49% YoY, and topped 50% of Baidu General business revenue for the first time (52%)

  • AI Cloud Infra revenue grew 79% YoY, with GPU cloud revenue accelerating to 184% YoY (from 143% the prior quarter)

  • Apollo Go delivered 3.2 million fully driverless rides in Q1, sustaining triple-digit YoY growth; over 22 million cumulative rides as of April

  • March daily average external token consumption on the MaaS platform reached nearly 7x the year-ago level; ERNIE assistant DAUs nearly doubled YoY

  • Operating cash flow stayed positive for a third consecutive quarter at RMB 2.7 billion; non-GAAP operating income of Baidu General business rose 39% QoQ to RMB 4.0 billion

Concerns

4
  • Total Baidu revenue of RMB 32.1 billion fell 2% QoQ; online marketing (IT) revenue of RMB 6.2 billion declined 8% QoQ

  • Cost of revenues rose 7% QoQ to RMB 19.6 billion on higher AI cloud costs

  • Total other income net fell to RMB 626 million from RMB 1.2 billion the prior quarter

  • Domestic AI chip supply remains relatively tight versus demand, constraining near-term cloud capacity

Guidance & targets

CategoryTargetConfidence
Cloud blended margin trajectory
structural improvement / long-term profitability
high materiality
Medium
Medium-to-long-term company operating margin
compelling and sustainable expansion
high materiality
Medium
Capex / AI investment intensity
maintain strategic investment intensity while preserving financial discipline
high materiality
Medium
Apollo Go robotaxi profitability
path to profitability becoming clear; overseas potential for much stronger profitability
high materiality
Medium
Shareholder returns (buyback + dividend)
continue buyback program and dividend policy; balance with AI investment
high materiality
Medium
Apollo Go international expansion
commence open-road testing in Switzerland; begin London testing with Uber and Lyft soon
medium materiality
Medium

Segment performance

SegmentRevenueYoYQoQMargin
Baidu General business
Returned to positive YoY growth after several quarters of decline; increase in 'others' driven by AI cloud growth.
Baidu Core AI-powered business revenue: RMB 13.6 billionAI-powered share of General business revenue: 52%Employees as of Mar 31, 2026: ~28,000
RMB 26.0 billion+2%flatNon-GAAP operating income RMB 4.0 billion (+39% QoQ)
Baidu Core AI-powered business
Crossed above half of General business revenue for the first time; driven by AI Cloud Infra and AI applications.
Share of Baidu General business revenue: 52% (first time majority)
RMB 13.6 billion (exceeded RMB 13 billion)+49%——
AI Cloud (total)
Combined AI Cloud Infra and AI applications; GPU cloud accelerated from 143% growth the prior quarter to 184% YoY.
AI Cloud Infra revenue growth: +79% YoYGPU cloud revenue growth: +184% YoY
RMB 11.3 billion———
Online marketing (IT)
Legacy marketing revenue declined QoQ; associated content and traffic acquisition costs decreased.
RMB 6.2 billion—-8%—

BIDU operating KPIs by quarter

BIDU operating KPIs stated on its earnings calls, by fiscal quarter
KPI Dec 2024 Q4 FY24 Jun 2025 Q2 FY25 Sep 2025 Q3 FY25 Dec 2025 Q4 FY25This call Mar 2026 Q1 FY26 Jun 2026 Q2 FY26Change vs prior quarter
Employees Baidu Core
~31K Baidu Core had approximately 31,000 employees as of December 31, 2024. Source: earnings call transcript
~31K Baidu Core had approximately 31,000 employees as of June 30, 2025. Source: earnings call transcript
~31K Baidu Core had approximately 31,000 employees as of September 30, 2025. Source: earnings call transcript
————
Monthly active users (MAU) Baidu App—
735M In June, Baidu App's MAU reached 735 million representing a 5% year-over-year growth. Source: earnings call transcript
708M In September, Baidu App MAU reached 708 million, up 1% year-over-year. Source: earnings call transcript
~700M During the Chinese New Year, we moved quickly to embrace the latest AI agent innovation by integrating OpenClaw, a recently popular open source agent framework directly into Baidu app with one-click access, enabling our users to immediately benefit from cutting-edge agentic AI capabilities with an MAU of around 700 million. Source: earnings call transcript
———
Employees Baidu General Business———
~29K Baidu General business had approximately 29,000 employees as of December 31, 2025. Source: earnings call transcript
~28K Baidu General business had approximately 28,000 employees as of March 31, 2026. Source: earnings call transcript
~27K Baidu General Business had approximately 27,000 employees as of June 30, 2026. Source: earnings call transcript
—

Operating figures the company stated on at least two of the calls shown, checked against each call's transcript. Click a figure to read the sentence. A dash means it was not stated that quarter.

Product announcements

ProductTypeDetails
ERNIE 5.1launch
Miaoda 3.0launch
Famou agent 2.0launch
Overseas digital human platformlaunch
DuMate (general-purpose agent)launch
Apollo Go standalone app (Dubai)launch
Kunlunxin inference model supportupdate

Deals & partnerships

Uber and Lyft partnership

First Apollo Go vehicles arrived in London in preparation for robotaxi testing with Uber and Lyft in the right-hand-drive UK market

Unity, Honor, Opel, Vivo customer contract

Named clients in Baidu AI Cloud's infrastructure client base; several prominent new clients including leading model companies added this quarter, with existing top-tier clients deepening collaboration

TikTok (and other China/overseas partners) partnership

Digital human partner base in China and overseas including TikTok and others, with several partners deepening collaboration on e-commerce live-streaming digital humans

Undisclosed partner (Hainan airport rental model) partnership

Partnered in Hainan on a rental model with fully driverless vehicles stationed at the arrival level of a Hainan airport, exploring new robotaxi monetization beyond ride-hailing; counterparty name [indiscernible] in transcript

Risks & headwinds

Online marketing (legacy IT) revenue decline Q1 FY26

IT revenue RMB 6.2 billion, -8% QoQ

Mitigation:Growth of AI-powered/AI cloud business offsetting the legacy decline; return to positive General business YoY growth

Domestic AI chip capacity and supply-chain constraints near-term

supply relatively tight; demand growing faster than supply (unquantified)

Mitigation:Actively expanding capacity and improving resource efficiency; Kunlunxin scaling and full-stack optimization; expects domestic supply capabilities to improve over time

Rising cost of revenues from AI cloud scaling Q1 FY26

Cost of revenues RMB 19.6 billion, +7% QoQ

Mitigation:Partially offset by lower content and traffic acquisition costs; mix shift toward higher-margin GPU cloud and server utilization improvements

Robotaxi still in investment phase / low domestic fares ongoing

UE breakeven only in Wuhan; still investment phase (unquantified overall losses)

Mitigation:Improving unit economics; more attractive overseas pricing environment expected to deliver stronger profitability as scale ramps

Robotaxi operational complexity at scale ongoing

unquantified — system and operational complexities emerging only at larger scale

Mitigation:Handling situations with rigor and using them to continuously strengthen operations, algorithms, and operational standards

Domestic AI chips trail global frontier in training ongoing

unquantified — still catching up with most advanced global products in certain frontier training scenarios

Mitigation:Focus on inference, where domestic chips are competitive; full-stack optimization and cost efficiency for Kunlunxin

Intense model competitive landscape ongoing

unquantified — active releases from players in China and globally

Mitigation:Continued conviction investment in ERNIE via application-driven approach; organizational adjustments to model teams; smaller/faster models optimized per scenario

Q&A highlights

What drove the AI cloud acceleration, is there sufficient compute capacity, and how does GPU cloud margin compare with traditional CPU cloud over the long term?

Dou Shen cited strong broad-based enterprise demand led by inference, MaaS traction, and full-stack efficiency. He said supply is relatively tight and Baidu is actively expanding capacity and improving resource efficiency. GPU cloud carries structurally higher margins than CPU cloud due to technical barriers, tight supply, and pricing power, plus Kunlunxin cost optimization, supporting durable blended-margin improvement.

“with demand remaining strong and the supply is relatively tight. We are actively expanding capacity and improving resource efficiency to better support the growing customer needs.”

asked by Alex Yao (JPMorgan) · answered by Dou Shen

3 min read 7 chapters

Detailed narrative

AI-powered business crosses the halfway mark

For the first time, Baidu Core AI-powered business exceeded 50% of Baidu General business revenue, reaching 52%, with revenue of RMB 13.6 billion, up 49% YoY. Management framed this as an important milestone confirming AI has become the majority of the revenue mix and the company's primary growth driver.

Combined AI Cloud Infra and AI applications drove total AI cloud revenue to RMB 11.3 billion in Q1. The shift is being positioned as both a growth and a margin story as higher-quality revenue streams scale.

AI Cloud Infra and GPU cloud acceleration

AI Cloud Infra revenue grew 79% YoY, well above the broader market, with GPU cloud revenue accelerating to 184% YoY from 143% the prior quarter. Management attributed momentum to surging demand across both training and inference workloads, with inference ramping especially fast, and to Baidu's full-stack advantage spanning proprietary infrastructure through applications.

GPU cloud carries structurally higher margins than traditional CPU cloud due to technical complexity, tight high-quality supply, and pricing power, supporting a durable blended-margin improvement. New clients added this quarter include leading model companies, Unity, Honor, Opel, and Vivo.

Kunlunxin self-developed AI chips

Kunlunxin is among the first domestic AI chips to achieve large-scale commercial deployment in a single AI computing cluster of over 30,000 accelerators, with industry-leading cluster performance and stability. It has been optimized and validated across the latest ERNIE and mainstream foundation models, with inference support recently extended to DeepSeek V4, GLM 5.1, and MiniMax M2.7.

Management positioned Kunlunxin as especially competitive in inference, where domestic chips are highly relevant, while acknowledging domestic chips still trail the most advanced global products in frontier training scenarios and face near-term capacity and supply-chain constraints.

MaaS platform and ERNIE foundation models

In March, daily average token consumption from external customers grew to nearly 7x the year-ago level, with MaaS revenue scaling rapidly. The model library was expanded beyond ERNIE to include DeepSeek, Gemini, MiniMax and others. Baidu recently launched ERNIE 5.1, delivering stronger capabilities, a more compact model size, and enhanced reasoning; it ranks first among Chinese models on the LM Arena text leaderboard and topped the LM Arena search leaderboard among Chinese models (fourth globally).

Management reaffirmed an application-driven iteration approach and noted organizational adjustments to its model teams.

AI applications portfolio

Management highlighted multiple AI-native and transformed products: a productivity agent showcased at Baidu Create for autonomous multi-step workflows across PC and mobile; digital humans with ~80% cost reduction over the past two quarters and a new overseas platform supporting 24 languages; Miaoda vibe-coding platform with MAUs up ~70% QoQ in March and domestic paying-user rate ~3x end-of-last-year levels (Miaoda 3.0 launched with enterprise and mobile versions); and Famou agent 2.0 for enterprise scenarios, deployed at Qingdao Port.

DuMate, a general-purpose consumer agent, was introduced as a new AI-native product form.

Apollo Go robotaxi scaling and internationalization

Apollo Go delivered 3.2 million fully driverless rides in Q1 with triple-digit YoY growth and over 22 million cumulative rides as of April. International expansion accelerated: on-track for open-road testing in Switzerland; first vehicles arrived in London for testing with Uber and Lyft; fully driverless operations across multiple zones in Dubai plus a standalone Apollo Go app launched there in late March; and a rental-model partnership in Hainan stationing driverless vehicles at an airport arrival level.

Management cited unit-economics breakeven in Wuhan and more attractive overseas pricing as the profitability path.

Financials and balance sheet

Total Baidu revenue was RMB 32.1 billion, down 2% QoQ; Baidu General business revenue was RMB 26.0 billion, up 2% YoY and flat QoQ; online marketing (IT) revenue was RMB 6.2 billion, down 8% QoQ. Operating income was RMB 3.2 billion (10% margin); non-GAAP operating income RMB 3.8 billion (12% margin).

Net income attributable to Baidu was RMB 3.4 billion (11% net margin), diluted EPS RMB 8.76 per ADS; non-GAAP net income RMB 4.3 billion, non-GAAP diluted EPS RMB 12.06 per ADS. Total cash and investments were RMB 279.3 billion, operating cash flow RMB 2.7 billion, and Baidu General business had approximately 28,000 employees as of March 31, 2026.

Summary of the company's earnings call; see how call summaries are prepared. Not investment advice.