Baidu's Profit Crashed 76% — And That's Good News for You

📉 Baidu just released their 2025 earnings. Net profit: down 76%. Let that number breathe for a second. Seventy-six percent.

But here's the thing nobody's talking about: for the first time in Baidu's history, they didn't even break out their ad revenue in the report. They just hid it. The company that built the "Chinese Google" model — ten blue links and an auction — stopped telling you how much money ads make.

That's not an accounting choice. That's a funeral.

And buried right next to it in the same report is something that should make every marketer sit up straight: AI revenue hit 40 billion RMB. Thirty-one percent of total revenue. Up from basically nothing three years ago.

So the old business is dying. The new business is exploding. And somewhere in between, there's a massive opportunity for anyone willing to pay attention.

🔎 What Actually Happened

Baidu's 2025 total revenue: 129.1 billion RMB. Sounds healthy, right? But their net profit cratered to 5.6 billion, down 76% year over year. And the ad business — historically the cash cow — is bleeding. Through the first three quarters of 2025, online marketing revenue sat around 47.5 billion, down roughly 14%.

Now here's where it gets interesting. Baidu's been saying "All in AI" since 2017. Almost a decade. They've poured close to 100 billion RMB into AI research, models, infrastructure. And in Q4 2025, AI revenue finally hit 43% of their core business income.

The bet is starting to pay off. But the transition is brutal.

🌊 The Two Waves That Killed Traditional Search Ads

This didn't happen overnight. There were two distinct waves.

🤔 Think about that from your chair as someone who buys Baidu ads. The entire mechanism that traditional PPC depends on — user types a keyword, sees your ad, clicks through — is being bypassed. The user goes straight to an AI that gives them a synthesized answer. No SERP. No ad slots. No click.

Wave one took your audience's attention. Wave two is taking the very interface your ads depend on.

🔍 So Is Search Dead?

No. And this is the part that matters.

Search — the human need to find answers, compare options, discover solutions — that's not going anywhere. It's as fundamental as asking a friend for a restaurant recommendation. The need doesn't disappear. The delivery mechanism changes.

Right now, when someone in China asks an AI assistant, "I need help running advertising campaigns in China as a foreign company, who should I work with?" — the AI doesn't return a list of ten blue links. It generates a response. A paragraph, maybe two. And in that paragraph, it names names. It recommends specific companies, specific services.

🎯 If your brand shows up in that generated answer, you win. If it doesn't, you're invisible. And no amount of traditional PPC spend changes that — because there's no ad slot to buy.

This is why the concept of 🎯 GEO — Generative Engine Optimization — matters so much right now.

You've heard of SEO: optimize your website so search engines rank it high. GEO is different. You're not trying to rank on a results page. You're trying to influence what the AI says when someone asks it a question. You want the AI to mention your brand, your service, your product — in its own words, in the answer it generates.

That requires a completely different playbook. Not bid management. Not quality score optimization. You need to build the kind of digital authority, the kind of consistent, structured, credible web presence, that AI models learn from and trust enough to surface in their answers.

📢 What This Means If You Advertise in China

1️⃣ PPC isn't dead, but its shelf life is shorter than you think. Baidu's own oCPC system already uses machine learning to auto-optimize bids. Traditional search ads will still generate leads in 2026, maybe 2027. But the margin is shrinking, the cost is rising, and the surface area where ads appear is contracting as AI answers take over more search queries.

Use PPC now. Seriously — it still works, and we help clients do it every day at baidumarketing.com. But don't build your entire China strategy on a channel that's visibly declining.

2️⃣ If you don't have Chinese-language content, the AI doesn't know you exist. Baidu's AI models are trained on Chinese-language data. If your brand has no presence on Zhihu (China's Quora), no mentions on Baidu Baike, no Chinese-language case studies, no reviews on Chinese platforms — the AI literally cannot recommend you. It's not choosing not to. It can't.

3️⃣ The early mover advantage is real and it's time-limited. Right now, GEO is wide open. Most Chinese companies haven't figured it out yet, let alone foreign brands. The company that starts building authoritative Chinese-language content today will have a compounding advantage that's nearly impossible to replicate later.

It's like SEO in 2005. The people who understood it early owned the next decade. The people who showed up in 2015 paid ten times more for the same results.

✅ Three Practical Steps

Step 1: Keep your PPC running, but start measuring AI visibility. Go to a tool like 27GEO.com or just manually ask Baidu's AI and a few other Chinese AI assistants questions about your industry. See if your brand comes up. Note who does get mentioned. That's your competitive benchmark, and it tells you what the AI currently considers authoritative in your space.

Step 2: Invest in Chinese-language content that answers real questions. Not translated press releases. Not English blog posts run through Google Translate. Real, useful, Chinese-language content that addresses the specific questions your target customers ask. Industry guides. Comparison articles. How-to content. Stuff that Zhihu users would upvote. Stuff that Baidu's AI would pull from.

Step 3: Build your entity presence on Chinese platforms. Baidu Baike page. Zhihu answers. Xiaohongshu presence if your product is consumer-facing. Mentions in Chinese industry publications. Every one of these is a signal that feeds into how Baidu's AI understands your brand.

None of this replaces your ad spend overnight. But it builds the foundation for the channel that's replacing traditional search ads.

📈 The Numbers Behind the Shift

Let's put some concrete numbers behind this. According to Baidu's 2025 annual report:

This is what a transition looks like from the inside. It's not a cliff — it's a slope. But slopes can be just as dangerous if you're not paying attention.

🕵️ What the Hidden Ad Revenue Actually Means

Here's the detail that got buried in the earnings report: Baidu stopped breaking out online marketing revenue as a separate line item. Previously, they'd show "online marketing services" as its own segment. Now it's folded into "Baidu Core" along with AI cloud and other initiatives.

Ask yourself: why would a company stop showing you the revenue breakdown for its historically largest business?

Two possibilities. Either they're reorganizing for the future and the old categories no longer make sense. Or they're hiding a number they don't want investors to focus on. In all likelihood, it's both. But the signal is clear: Baidu doesn't want you watching their search ad business decline in real-time.

For advertisers, this has practical implications. When you negotiate contracts with Baidu, when you set budgets, when you measure ROI — understand that you're operating in a business that's losing strategic importance to its own parent company. The attention, the innovation, the pricing incentives are all shifting toward AI. Your search ads are becoming legacy infrastructure.

🔮 Reading the Tea Leaves on AI Monetization

Baidu's AI revenue hit 40 billion RMB in 2025. What is that actually comprised of? Industry analysis suggests:

• AI cloud services: Enterprise customers paying for access to Ernie Bot APIs, model fine-tuning, and infrastructure — roughly 55% of AI revenue
• AI-enhanced search features: Premium search capabilities powered by AI, charged to enterprise clients — roughly 25%
• AI assistant monetization: Early experiments with sponsored answers in Wenxin (Baidu's ChatGPT equivalent) — roughly 20%, but growing fastest

That last line is what matters for your planning. Baidu is already experimenting with sponsored AI answers. Not traditional ads — actual mentions within AI-generated responses that are paid placements. The format is still being refined, but the direction is set.

When sponsored AI answers become mainstream, the entire optimization paradigm shifts. Instead of bidding on keywords, you'll be bidding on topics or queries. Instead of optimizing for CTR on a SERP, you'll be optimizing for inclusion in AI-generated responses. The measurement changes. The KPIs change. The skill requirements change.

🌐 The Competitive Landscape Beyond Baidu

Baidu's struggle is happening within a broader competitive context that's worth understanding.

ByteDance's Doubao has emerged as the unexpected leader in AI assistant usage. The Spring Festival Gala sponsorship wasn't just brand awareness — it was behavioral conditioning. They taught hundreds of millions of users to open Doubao as their first AI interaction. Doubao's user base grew from roughly 100 million monthly active users in early 2025 to over 300 million by year-end.

Alibaba's Tongyi Qianwen (通义千问) maintains strong enterprise penetration, particularly among Alibaba's existing customer base. For advertisers, this matters: if your potential customers are using enterprise tools that are integrated with Tongyi, they may encounter AI-generated recommendations within their work tools, not just consumer apps.

Minimax, Moonshot (Kimi), and DeepSeek have carved out niches — Kimi particularly among younger users and researchers, DeepSeek among developers and technical users. Each has different content partnerships, different recommendation patterns, different advertiser opportunities.

The key insight: there is no single AI search interface that dominates. Chinese users are distributed across multiple platforms. Your GEO strategy needs to account for this fragmentation.

🏆 The Foreign Brand Advantage

Here's something counterintuitive: foreign brands may actually have an easier entry point for GEO than domestic Chinese brands in certain categories.

The reason: most Chinese AI models were trained on massive corpora of Chinese-language text. Foreign brands with limited Chinese content presence are essentially invisible. But for brands that invest in building that presence — particularly brands with strong credentials, international certifications, case studies, and thought leadership — the AI models have less noise to work through.

Think of it this way: in many B2B industrial categories, there are a dozen dominant domestic Chinese brands with enormous training data presence. A foreign entrant might struggle to compete on pure mention volume. But if that foreign brand establishes itself as the authoritative "international option" in its category — the one that gets cited in comparisons, the one that appears in expert discussions about global best practices — it can own a meaningful position in AI-generated recommendations without needing to match the domestic brands' content volume.

🕵️ Practical GEO Audit: Where Do You Stand?

Before you can build a GEO strategy, you need to understand your current position. Here's a practical audit framework:

⏰ The Transition Timeline

How fast is this happening? Based on current trajectories:

💰 Building Your Transition Plan

Here's a framework for thinking about how to allocate attention and budget across the transition:

🎯 The Bottom Line

Baidu's 76% profit crash is not a failure story. It's a transition story. The company that built the world's second-largest search engine by user count is navigating one of the most disruptive technology shifts in the history of information retrieval. They're making the right long-term bets. The short-term pain is real.

📢 For advertisers, the message is: the channel you're using today is declining, but the replacement channel is still being built. You have a window — maybe two years, maybe three — to position yourself in the AI-native world before the competitive dynamics calcify.

🧠 The GEO playbook is different from the PPC playbook. It's slower to show results. It requires different skills, different content strategies, different relationships. But the companies that figure it out now will own a significant advantage when the transition completes.

💡 Fifty billion RMB changed how a billion people find information. That change is not reversing. The only question is whether your brand will be there when they ask.