--- title: "Search no longer valued by the market: Baidu should split off its AI business without delay" type: "Topics" locale: "en" url: "https://longbridge.com/en/topics/43484396.md" description: "While overseas valuations for Anthropic are soaring far ahead, and domestic players like Minimax and Zhipu have also created modest wealth myths, many naturally assume that any company linked to AI is guaranteed a solid valuation. The reality, however, is filled with counterexamples, with Baidu being the most representative. In early 2023, following the explosive popularity of ChatGPT, Baidu quickly released its Ernie Bot large model, briefly leading China's LLM development. Three years later, Baidu's AI-related revenue now accounts for over 50% of its total, effectively transforming it into an AI-driven company on paper..." datetime: "2026-08-19T11:38:16.000Z" locales: - [en](https://longbridge.com/en/topics/43484396.md) - [zh-CN](https://longbridge.com/zh-CN/topics/43484396.md) - [zh-HK](https://longbridge.com/zh-HK/topics/43484396.md) author: "[仝志斌](https://longbridge.com/en/profiles/10247393.md)" generator: "portal-rs" --- # Search no longer valued by the market: Baidu should split off its AI business without delay ![图片](https://pub.pbkrs.com/uploads/2026/b8f6be2ac32d2d88803a2a0d131e6f5e?x-oss-process=style/lg) While Anthropic's valuation abroad has surged ahead, and domestic players like Minimax and Zhipu have created modest wealth myths, many naturally assume that any company linked to AI is guaranteed a solid valuation. The reality, however, is filled with counterexamples, the most representative being Baidu. Following the explosion of ChatGPT in early 2023, Baidu quickly released its Ernie Bot large model, briefly leading China's LLM development. Three years later, while AI-related revenue accounts for over 50% of Baidu's total—formally transforming it into an AI-driven company—its valuation has not risen but fallen, with the stock price dropping nearly half from early 2023. In the same AI track, where exactly does Baidu fall short? The core arguments of this article are: First, the decline of Baidu's AI business stems from sluggish capital expenditure (CapEx), caused by insufficient financing capabilities, rooted ultimately in bundling search and AI businesses together; Second, to protect short-term financial metrics, Baidu cut CapEx, and the increased spending in 2026 may severely impact full-year profitability; Third, the search business no longer commands significant valuation power in the market, and Baidu should sever ties decisively. Bundling Search and AI: Dragging Down Baidu's Compute Reserves In mid-2023, we published an article suggesting Baidu spin off its AI business to enhance financing capabilities and secure crucial compute resources. To date, Baidu has not adopted this strategy, which we believe is the root cause of subsequent controversies. We know Baidu relies on search as its foundation, with online advertising as its primary business model. Since the rise of mobile internet, especially short video, advertisers have shifted their budgets from search to mobile platforms, particularly short video. This industry and product trend is inevitable for Baidu. With the launch of the LLM track, Baidu effectively formed two major business segments: A: Traditional search business, low growth, high gross margin, unclear prospects, and diminishing valuation premium; B: LLM-led business, bright prospects, but facing short-term losses and requiring substantial long-term CapEx. If these two segments are bundled for financing (via equity or debt issuance), **investors, even those who appreciate Baidu's LLM business, would be forced to buy into the declining search business, which is hard to accept.** Thus, we see Baidu's CapEx stagnation in recent years. ![图片](https://pub.pbkrs.com/uploads/2026/548ca72fbeb27e3eac50dfe7c5268af3?x-oss-process=style/lg) Since 2023, Baidu's CapEx scale has remained largely flat. We also know that LLMs are extremely capital-intensive, requiring massive compute centers for training and R&D. Without sufficient compute backing, LLM companies struggle to compete. The viral success of 'Lobsters' in 2026 pushed industry-wide profitability forward, prompting almost all enterprises to increase CapEx. Tencent, for instance, raised its Q2 CapEx to up to ¥58bn. Baidu also showed a clear shift, reaching ¥11.4bn in the period, a qualitative leap from ¥3.8bn YoY. However, even with this effort, Baidu's spending remains a drop in the bucket compared to Tencent. Is Baidu unwilling to invest? No. **As mentioned earlier, bundling high-quality assets with poor ones weakens the company's financing ability.** Many companies, including Tencent and Google, have issued medium-term bonds recently. What about Baidu? 2025 was a heavy bond year for Baidu: issuing ¥14.4bn RMB bonds + $2bn USD exchangeable bonds, mainly for "refinancing" to optimize debt structure and reserve funds for AI compute expansion. There are no financing plans for 2026. ![图片](https://pub.pbkrs.com/uploads/2026/60fa42fbbaee5cb4f06745bb3258f9c8?x-oss-process=style/lg) The root cause lies in ROIC (Return on Invested Capital), the key metric for evaluating capital efficiency. In 2024, this indicator rose to 7%, a recent high, reigniting investor confidence and opening financing channels. Post-2025 earnings, however, the metric dropped to around 1%. Such a low return cannot cover costs, causing investors to retreat. This clarifies why Baidu's LLM development lags peers and CapEx remains stagnant: First, bundling good and bad assets prevents separate financing for the LLM business, forcing investors to rely on consolidated ROIC; Second, to protect ROIC, Baidu had to cut CapEx short-term (reducing amortization costs), leading to conservative spending from 2023-2025 and lagging compute reserves; Third, increasing CapEx in 2026 will inevitably pressure the P&L statement, creating significant full-year profit pressure and further lowering ROIC below market expectations, tightening financing again. To summarize: The weakness of the search business has constrained the entire group. Baidu should decide to spin off its AI business to boost independent financing and reassure investors. **Search is no longer valued by the market** Some may still question: Online marketing remains Baidu's core profit and cash driver; splitting it might be risky. I won't dwell on basic concepts like "consolidated financial statements" or "AI funding needs far exceed search contributions," but offer another perspective: Search's valuation in capital markets is near zero, or even negative (dragging down AI valuation). In Q2 2026, Baidu's core AI new business generated ¥12.5bn in revenue, implying an ARR of roughly ¥50bn. Let's estimate the AI segment's valuation separately: Intelligent Cloud ARR is ~¥30bn. With a ~5x PS multiple, the valuation is ¥150bn; AI Applications & Native Marketing ARR totals ¥20bn. With a market PS of >10x, the valuation is at least ¥200bn. **Even using conservative multiples, Baidu's AI segment should be valued at over ¥250bn.** Yet Baidu's total market cap is only HK$250bn (~¥230bn). This proves: 1) Bundling good and bad assets results in 1+1<2; search drags down AI valuation; 2) Spinning off benefits the enterprise. When pondering why Baidu hasn't spun off, friends suggested looking at internal management: post-split, many search employees (including mid-to-senior management) would miss out on AI equity premiums, becoming obstacles to strategy. Consider this perspective. However, a reminder to Baidu's management: the time window is closing; delay is not an option. $Baidu(BIDU.US) ### Related Stocks - [09888.HK](https://longbridge.com/en/quote/09888.HK.md) - [BIDU.US](https://longbridge.com/en/quote/BIDU.US.md) - [89888.HK](https://longbridge.com/en/quote/89888.HK.md) - [HBUD.SG](https://longbridge.com/en/quote/HBUD.SG.md) - [02513.HK](https://longbridge.com/en/quote/02513.HK.md) - [00700.HK](https://longbridge.com/en/quote/00700.HK.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [00100.HK](https://longbridge.com/en/quote/00100.HK.md) - [ANTH.NA](https://longbridge.com/en/quote/ANTH.NA.md) ## Comments (1) - **Susie7 · 2026-08-19T12:00:42.000Z**: Baidu's valuation is so low, is it worth buying? --- > **Disclaimer: This article is for reference only and does not constitute any investment advice.**