Baidu CFO Haijian He on AI Revenue, Robotaxis
THE SO WHAT
When Baidu says 50% of revenue is now AI-related, it’s telling you the Chinese internet playbook has already rotated from ads to infrastructure and autonomy. If you’re competing in that market, assume your rival’s P&L is subsidizing long-horizon bets like robotaxis with core AI services — not the other way around.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIThe Trump White House is ready to regulate AI, but it's exactly the wrong body to do so, and its control could become a problem
Frontier AI is now explicitly a federal power question — an executive order claiming “unprecedented control” over future models turns model deployment into a political dependency. If your roadmap assumes neutral, predictable US oversight, you’re already off; build for jurisdictional diversification and regulatory failover now.
Applied AISensor Tower: ChatGPT has become the fastest app to hit 1B global MAUs by far; ChatGPT's MAUs are up 62% YoY in Q2 to date, Claude's MAUs are up 640% YoY to 56M (Harshita Mary Varghese/Reuters)
1B MAUs for ChatGPT and 640% YoY growth for Claude mean the assistant layer is no longer a niche tool — it’s a consumer platform with real distribution gravity. If your product strategy assumes users will start in your app, update it — they’re starting in an AI interface and pulling you in via plugins and APIs, or not at all.
Applied AIBillionaire Ambani’s Jiostar Platform Bets Big on All-AI Series
A major streamer betting on fully AI-generated series — after a profitable 2,500-year-old war epic remake — means content cost curves are about to diverge hard between those who automate and those who don’t. Studios and brands need to decide now whether their moat is IP, talent, or throughput, because AI-native players are optimizing only for the last one.
Applied AISource: Google asked Google Play app devs to join a "confidential content offer pilot", offering to pay for access to codebases to use them to train AI tools (Jason Koebler/404 Media)
Training data is now a line item — Google paying Android devs for private codebases turns your repo into an asset class. If you’re not explicitly pricing and licensing your code and logs for model training, you’re giving away the most valuable part of your stack for free.