Perhaps less reliable accounting, political risk, sometimes risky diversification, a half-baked shift to mobile, and profitability compressed by heavy investments in R&D easily justify this gap.

The upturn of 2020 - from which MarketScreener, which at that time had the stock among its US portfolio positions, had benefited very well - will have been short-lived. As shown by the quarterly results published yesterday, revenue growth is modest and profitability is still not expanding.

This situation is unlikely to improve over the next few quarters, as Baidu has to work hard to catch up in the field of artificial intelligence. A few weeks ago, the demonstration of Ernie - its alternative to ChatGPT, Bard and others - was a flop.

However, if there is something to be welcomed, it is the restructuring of iQIYI - the "Chinese Netflix" - whose accounts have obviously been put back in the green after a series of loss-making years. On this point, Baidu is doing much better than its American peers.

On a consolidated basis, Baidu should be able to generate between $2 and $2.5 billion in free cash flow this year. The current enterprise value of $36 billion therefore represents a multiple of between x14 and x18 earnings. This is a reasonable valuation, but it does not leave much room for error.

Some would argue that the sum of the parts is far greater than the enterprise value, which would not necessarily do justice to Baidu's leadership position in autonomous driving technologies. There's no need here to detail the opportunity that such a lead can represent in a market as colossal as the Chinese market.

As we recalled a few days ago with the example of News Corp, the sum-of-the-parts exercise, although elegant, rarely leads to satisfactory investment results.