Waymo, Alphabet's autonomous driving unit, raised capital this year at a $126bn valuation. Baidu's entire business, search engine, cloud arm, large language model and a robotaxi fleet of its own, is valued at around $38bn. That's not a typo, and it's not really a story about self-driving cars. It's a story about where investors have decided to look for value, and where they haven't.
The mismatch
Baidu's Apollo Go runs a comparable number of rides each month to Waymo. Alongside that, Baidu operates China's largest search engine, Ernie Bot (a large language model built to handle complex text and visual prompts), and Baidu AI Cloud, which sells storage, compute and enterprise data tools. Waymo is a single business line inside Alphabet. Baidu is four businesses wrapped into one, and the market prices the four together at less than a third of the one.
Why the gap exists
Part of the answer is structural, not fundamental. Roughly 80% of the MSCI China index is made up of Hong Kong and US-listed companies. Only 15.6% consists of companies listed on the Shanghai and Shenzhen exchanges, which is where China's semiconductor and advanced technology ecosystem is actually concentrated. Samsung, SK Hynix and TSMC alone account for roughly a third of the MSCI Emerging Markets index. Alibaba and Tencent together represent about 4%, despite building and deploying AI at a scale comparable to their US counterparts.
The Hang Seng Tech Index has been recovering, Alibaba and Tencent both posted strong single-day gains in the first week of August on AI-related news, but the index spent much of the past four years trading near its lowest levels since late 2022. A rebound off a multi-year low is not the same as a re-rating, and the underlying discount to global tech peers hasn't closed nearly as fast as the share price moves suggest.
The mispricing isn't in the technology. It's in where investors are choosing to look for it.
What it means for portfolios
Index-level exposure to "China tech" mostly buys Hong Kong and US-listed names that already trade at a premium to their Shanghai and Shenzhen-listed peers. The companies building and deploying AI at scale inside China, the ones a search engine, a foundation model and a cloud platform actually depend on, sit largely outside the benchmarks most portfolios track. Capturing that gap requires selecting individual names rather than buying the index, which is a different, more active decision than most allocators are used to making in this part of the market.
None of this requires a view on whether Baidu's robotaxi business ultimately succeeds against Waymo. It only requires noticing that the market is charging a steep premium for one version of the same bet and a discount for the other, and asking whether that gap is about risk, or simply about where the index provider drew the line.
Happy to send over a sample if it would be useful.