Baidu (BIDU): Valuation Update

It's been an eventful few months for Baidu's stock since my initial writeup on March 13th:

The stock price was hammered due to fears of 1) "core" business (search) growth declining 2) general China macroeconomic weakness, which is impacting advertising spend and 3) trade talks between the USA and China falling apart, worst case escalating into a complete severance of economic ties.

Some of these fears were put to rest with their Q2 results, but pessimism remains. The broader economic and political threats are also largely out of Baidu's control, so I will focus on analyzing the "core" growth threats.

Core Growth Threats

Baidu "core" (which excludes iQiyi) makes the majority of its revenue from search advertising. Historically, traffic to their search has come from Baidu's desktop / mobile websites. However, more and more traffic has shifted to native mobile "apps", resulting in a few major threats:

  1. Baidu is not able to "catalog" proprietary content that is being created in the new "walled gardens" of these native apps, as opposed to the "open" web. This is resulting in a greater share of "unsearchable" content.
  2. Competitors are pushing all-in-one "super-apps" that incorporate search, bypassing the need to go to Baidu directly.
  3. New popular apps from competitors like Bytedance are increasing ad supply and driving down ad prices

Threat: Walled Gardens

The main consequence of increasingly siloed native app content will likely be that Baidu will need to spend more money to access content, or create its own. This will probably take the form of: external investments, like Baidu's recent one into Zhihu (a Q&A type site); outright acquisitions; and investments into internal content farm platforms like Baijiahao.

Short term, this will likely drive up operating costs and erode margins. Long term, it's hard to see where things will end up. There has been a lot of criticism of Baidu's increasing propensity to link to its own content. However, a quick check shows that competing search engines (Shenma, Sogou) are no better. Amidst all of the fear-mongering and hyperbole ("Baidu is dead!"), it is also worthwhile to note that their market share has been relatively stable:

Market-share based on my own internal calculations using SimilarWeb data, which seems more accurate and far less volatile than the oft-cited statcounter data.

Even in a world where every Chinese internet titan jealously hoards and guards its own content, I don't see search going away completely. Users will just need to search multiple sources, and companies that focus on returning the highest quality content and most accurate results within their own internal ecosystems will likely continue to prosper. Notably, Baidu is making a push into more "structured" content - either through "mini-apps" or "managed" webpages - likely in an effort to make it easier to automatically filter good content from bad using AI.

This will be a tough problem to solve completely, especially considering the lack of goodwill that the company seems to have among the Chinese public, but management seems aware of the challenge and heading in the right direction. Recent evidence of this is their stated willingness to ban lower quality medical advertisers from their new healthcare platform, despite the short term revenue hit.

Threat: Superapps

The trend / hype has been towards "all-in-one" super apps that can do everything. Baidu has been heavily criticized for being late to the app party, and part of the reason they missed profit expectations in Q1 2019 so widely is that they spent hundreds of millions of dollars promoting their app during the Chinese New Year Gala.

I don't really have any idea how Chinese internet user behavior will pan out, but a few key things to note are: Baidu's app is growing robustly (up 27% YoY in June to 188MM daily active users); they are driving "vendor" lock-in by prioritizing proprietary "managed" pages and "mini-apps" within their search results; and they claim that this app growth is sustainable at positive ROI's.

The app's continued popularity in China, well after the one-time branding binge early this year, seems to support these claims:

Source: App Annie; iPhone, China data.

Again, I see the most likely scenario as users accessing different, walled-off information using multiple super-apps. As long as Baidu can provide value with their own app, they should have a seat at the table.

Threat: Competition

ByteDance is commonly hailed as the world's most valuable startup ($75B), and it has recently launched a search engine accessible within its apps. The popularity of its apps is also adding to an oversupply of ad inventory, and advertisers that previously might have spent money with Baidu are allocating more to ByteDance.

While it would be a mistake to underestimate ByteDance, they face many of the same issues as Baidu in building out a search engine, like pay-to-play access to content. The main unique content that they do offer (entertaining, short form videos) is more likely to be consumed by users in a "experiential" / "discoverable" format than a "search-driven" format (one of their main value propositions is that they are good are proactive at recommending addictive content, users aren't "searching" for it and probably wouldn't even know "what" to search for).

ByteDance is also spending $1B+ a year on advertising and is still unprofitable, so it's unclear how sustainable their strategy is long term. It is also likely exponentially harder to create a good search engine than it is to curate a few thousand news sites, so their costs are probably not going to diminish.

More generally, many people seem to be making the mistake of thinking that all ad inventory is equal. Even if ByteDance is unleashing hoards of new ad inventory into the market, that doesn't mean it's any good. Ultimately, advertising spend will flow to places where it generates the highest return. It seems unlikely that generically targeted ads placed alongside videos of teenagers dancing, in a network battling with a reputation for sexual predation, are going to command high premiums, nor generate large returns for advertisers. They also lack the user-specific, personal data that a network like Facebook has. I suspect that part of the reason they are pushing into search is that they are trying to leverage their traffic into something with a more profitable business model.

That's not to say they won't find a way to be successful - they are clearly a remarkable company - but the fears are likely exaggerated.


While the search business faces headwinds, they don't seem like existential threats. And even if growth slows, or beings to decline, it should largely be offset by growth in their Cloud services division, which is on pace to generate $1B in revenue this year (see my original post for more discussion of this opportunity, which I believe is being greatly underestimated).

Baidu is also making progress with DuerOS, a voice assistant (similar to Alexa, Siri). In Q2 2019 they shipped ~4.7MM smart speakers. Margins and revenue are likely small at this point (one of the speakers sells for only $14 USD), but even at these relatively low ASPs the revenue is starting to become material.

Their self-driving car platform (ApolloOS) also seems to be maintaining a lead in China, but to be conservative I'm not assigning this much value to it yet ($1B; for comparison, Waymo, the clear leader, is valued anywhere from $35B - $175B).

I'm also not counting on any incremental profit from DueroOS or ApolloOS - but I do think that the speech / voice / driving data they are collecting will be funneled into improving their Cloud products (i.e. voice recognition, language translations, image recognition, etc), and will indirectly contribute in this way.

Updated Valuation

With that general context out of the way, here is my updated valuation of the company ($165, down from $208 in March):

Despite the adjustment, and despite the fact that the stock price has rebounded +12% to $109 in the past few days, it still seems considerably undervalued to me.

The key new assumptions are:

  • 25% chance of "catastrophic failure", defined as a "fire-sale" of sorts where the company liquidates shares for 50% of fair value. The probability of failure is driven by 1) potential delisting of Chinese stocks and 2) international condemnation (i.e. sanctions) from a potential heavy-handed crackdown in Hong Kong. This probability is fairly arbitrary, and I think that these events are still unlikely, but I felt like I needed to capture some of the existential, macroeconomic risk. It should be noted that this assumption, and not anything related to the factors within Baidu's control, is responsible for ~50% of the adjustment in fair value.
  • Decreased revenue growth in the search advertising business, mostly offset by increases in hardware sales and cloud growth
  • Decreased operating margin in the core business, largely as a result of the lower margin product mix described above

* Note that iQiyi (a streaming video service similar to Netflix, in which Baidu has a 56.7% stake) is now valued using a separate DCF model. This allows me to get a little more granular with the operating characteristics of each business line and should hopefully yield a more accurate long term valuation. I'm valuing IQ at a significantly lower value ($5) than its current stock price ($18).

Disclosure: I am long BIDU

Show Comments