$Baidu(BIDU)$  $BIDU-SW(09888)$  After going through BAIDU Q1'25 earning release, several key points I'd like to note the following:

1) It's profitable ad business is dropping -6% yoy- driven by consistently shrinking Managed Pages rev (50% down to 47% as a proportion of ad-business yoy); it's widely known that BIDU content ecosystem suffered from walled gardens of popular social media app eg. WeChat, Douyin, Xiaohongshu due to preference & convenience of updating information via image & video format of Chinese people.

In a way, BIDU has been complacent in stimulating stakeholders engagement & going into multimodality of search despite already possessing the tech to do so for a long time (they were first to develop machine vision & machine learning in China). 

In addition, with recent opening up of China, open visa tourism were introduced along with multiple business friendly gestures. Foreigners are just starting to get to know China again. This development is likely to be a long-term multi-year bullish trend boosting tourism to China. Base on result, they did not capitalize on foreigners affinity to Google in getting to know China by adapting Baidu App to foreigner (eg. English translation or option on opening BIDU app) nor did they partner up with tourism company to recommend Baidu app as a suite of essential app in China. With increase in users come traffic & demand alongside the need for businesses & stakeholders to update their business info & marketing in BIDU ecosystem. 

The only positive metric is reacceleration of  MAU to +7% qoq sequentially. With increase in MAU, should management make the right moves in targeting tourism industry, I believe ad business should rebound in contrary to what most analyst thinking because businesses would always go wherever there are users. That's one business law that will never change.

2) Non-ad business is the bright spot growing at+40% yoy driven by acceleration to +42% yoy in AI Cloud business which accounts for 71% of non-ad business. This is where management has been executing in the enterprise AI cloud segment by providing essential support to developers & enterprises rendering affordable application to scales via distillation of database (Qianfan) of open-sourced LLM models. In addition, they've also been providing AI agents to enterprises & digital human services to e-commerce platforms. (Now we know who provides AI solutions to e-commerce platforms).

While entire world was competing on outbeating each other on foundational models benchmark, Robin is clearly aware that value of LLM models lies in the application of it in real lives. Currently, AI Cloud is mostly subscription base with some project based revenue with adj. operating income (most likely management metric*) in the teens.

Taking corresponding Q1'2024 as baseline, a quick calculation tells us that BIDU has incur additional +RMB 1.892B in COGS & SGA Exp to bring in additional +RMB 1.982B in AI cloud revenues. That's a Rev:Exp ratio of 1.047 meaning that for every additional 1 RMB they spent, they're only able to make ~RMB1.05. This tells you where China's AI cloud industry is at at the current development phase.

If ERNIE is able to execute on this front, a further acceleration in spending on AI cloud with increase cloud adoption would at the minimum sustained the current growth rate or even further accelerating growth rates for a multi-year period with increasing adj. operating income margin. In that case, revaluation of BIDU would surely come to reflect it's promising prospects at the current record low Chinese interest rates & lost of credibility of the USD.

3) Last promising growth arm, ApolloGo which reported rides growing at a +75% yoy (1.4m rides) in the quarter with cumulative rides exceeding 11m. Right from the start, BIDU has developed it's own robotaxi software solution & open-source it through Apollo Scape so that others could reduce R&D cost by using their codes which contributed to a bustling robotaxi industry in China only limited by regulator's good will. This along with BIDU's AI efforts likely draws gov's favor to BIDU as it birth an ecosystem of high tech industry development.

Currently, they're charging RMB 4 - RMB 16 per 10km ride with their latest RT6 robotaxi costing below USD30K each. Assuming a fleet of 1000 robotaxis (BIDU guidance from last year*) clocking 1.4m rides in Q1'25, each robotaxi makes abt 15 rides per day (similar to Pony.AI's #). If each ride goes for RMB16 (10KM), it implies a RMB 22.4M rev in Q1.

At current robotaxi fare, ApolloGo would take 2.5 years to recoup cost of RT6 construction without taking into account COGS & othr operating expectations.

Drawbacks were slow waiting time when hailing for one (can be fixed by AI optimized algorithm & increase in # of robotaxis), following a longer route base on its map (also can be fixed by it's AI algorithm) & biggest challenges of all, pick up in robotaxi expansion draws ire of local cab drivers who depends on it for a living. Advantages were absence of drivers which allow a sense of freedom & a more relax environment for users & extremely safe driving style. Judging from the earnings call, management is fully focused on AI t& therefore, robotaxi will take a back seat for now due to heavy CAPEX, regulation & social ramification.

Lastly, BIDU's earnings result benefited from increase in fair value gain & long-term investment of RMB 2+B which are non-operational in nature. I continue to see a favourable benefit-risk ratio towards the upside for BIDU over the long term. Cheers.