Despite being a full-stack AI player, Baidu's CFO identifies the cloud as the most critical layer. It serves as the central platform for deploying not only their own model (Ernie) but also third-party models, making it the key to monetization, inference deployment, and overall ecosystem control.
Baidu forgoes rigid policies allocating AI compute 'tokens' to employees based on seniority or title. The CFO argues the unit cost of compute drops so fast that such policies become obsolete in weeks. They prefer empowering talent with ample resources, trusting them to prioritize tasks efficiently in a nimble environment.
To compete for top engineers, Baidu's pitch emphasizes employee autonomy and access to its entire AI stack—from chips to applications. They encourage a 'one-person team' culture where individuals leverage internal AI agents to handle complex tasks, offering a unique value proposition of trust and end-to-end project ownership.
Baidu's rationale for developing its own silicon isn't to control the supply chain or dominate pre-training. It's a strategic focus on the AI inference market, which the CFO states accounts for 80% of incremental compute demand. Their chips are optimized for this specific task, creating a positive network effect with their cloud business.
The CFO frames balancing high growth, investment density, and shareholder returns as an 'impossible triangle.' Baidu navigates this by meticulously analyzing the full cash-back lifecycle (e.g., 20-40 months) for every dollar spent on AI. This allows responsible investment without sacrificing ambition or financial stability.
In contrast to the 'AI psychosis' of some US labs, Baidu’s CFO frames AI alignment as a technical challenge of robustness and data sanity. He suggests these issues are being efficiently addressed by a 'very collegial' global open-source community, indicating a more pragmatic and less alarmist approach to AI risk management.
Baidu's CFO admits China's low public cloud penetration (~30% vs. 90% in the US) and siloed data ecosystems have been a disadvantage. He argues that intense demand from AI models and agents is now forcing a rapid migration to the public cloud, which will help resolve this data fragmentation issue over time.
The key to mass robo-taxi adoption is economics, not just technology. Baidu's CFO identifies 60-80 cents per mile as the critical price point where using a robo-taxi becomes cheaper than personal car ownership in the U.S. The entire industry is racing to drive costs below this threshold to alter consumer behavior.
Baidu is replacing the classic internet metric of Daily Active Users with Daily Active Agents. This signifies a fundamental shift from monetizing user attention to monetizing task completion. The DAA metric aligns with their new business models, like result-driven payments and profit-sharing with enterprise clients.
Baidu's CFO observes that enterprise AI sales are no longer IT-centric discussions with CTOs about cost centers. Because AI agents can directly impact P&L (e.g., optimizing a shipping port), the primary sales conversation now happens with the CEO, who sees it as a strategic, top-down initiative with a clear budget.