Baidu (BIDU) may be one of the rarest of sightings: an undervalued Chinese tech stock.
It's traditionally hailed as China's Google, and operates in similar sectors:
- Search - With roughly 75% of China's market share
- Video - Both user generated content (like YouTube) and high-quality original programming (like Netflix)
- Cloud Services - Primarily focused on AI as a Service (AIaaS)
Unfortunately, even though it is a fabled member of the BAT triumvirate, its returns have greatly lagged its rivals and the market in general...
...primarily due to its slower growth and perceived risk factors.
Google operated in China until 2010, and enjoyed decent market share, but ultimately pulled out due to hacking and censorship concerns. Recently, there have been rumors of Google's return to China, but these rumors have been denied by the CEO.
I think Google is unlikely to return for a few reasons. First, they face significant internal resistance from employees, who are becoming more politically vocal and have already demonstrated their ability to impact commercial prospects. Second, Google has bigger concerns on its plate, like increased antitrust scrutiny, and is unlikely to do anything near-term (like partner in censorship with China) to upset its political positioning.
Short term, the only way I see Google returning is if China loosens the restrictions on its great firewall, an unlikely prospect.
Even if they do return, they will face a very different competitive landscape than previously, and there is no guarantee that they will enjoy similar success. Most notably, there has been a large shift towards native app usage in China, which will reduce Google's monopolistic ability to leverage its mobile browser to drive traffic to search.
App vs Browser
Baidu is much less dominant than Google is in search, partly because its competitors have invested heavily in alternatives, but also because Chinese consumers rely more exclusively on native mobile apps.
In a recent earnings call, the CEO, Robin Li, said that:
We believe the future of mobile search will shift toward closed-loop, native-app experience from the current search experience that directs users to HTML5 sites.
In other words, consumers are skipping the mobile browser (i.e. Chrome, Safari, etc) in favor of searching within apps directly. For example, imagine starting all of your travel searches with the Booking.com app, or all of your shopping searches with Amazon app, and all of your video searches with the YouTube app, etc - rather than with Google's mobile web search engine.
Furthermore, each Chinese internet giant is trying to turn their app into a "super-app" - basically, an app that runs other, mini-apps - in an effort to keep users dependent on their "super-app" and individual, closed ecosystems.
This has strong implications for branding, marketing and growth, which are reflected in the CFO's comments:
Our focus to place greater reliance on organic traffic growth means that we are shifting our financial model to spending upfront marketing dollars with returns spread out over the life of the users...
Our traffic-acquisition mix, shifting from TAC traffic to organic traffic, will dampen our profit margin in the near future, but with extensive internal ROI analysis, we believe over the long term, this will strengthen Baidu's foundation in search and feed, especially with users moving toward super apps and the increase in popularity of mini programs in China.
The way I interpret these comments are that Baidu (and Tencent and Alibaba...) are investing more heavily in brand marketing, which may not have an immediate, measurable return. They are also going to continue investing heavily in tangential, content-rich apps (i.e. short form video), with the goal of keeping this content exclusive to their network, in an arms race to make their "super-apps" and closed-content ecosystems stickier for consumers. This is not necessarily an existential risk, but it will drive up costs for everyone long term.
Brand / Quality
In my opinion, this is probably the biggest existential risk for Baidu.
The Baidu brand is not well-loved in China, ranking 125th out of 280 companies. Their search results page is viewed as overly commercialized and their ads as untrustworthy, most starkly exemplified in a widely publicized scandal in 2016 that resulted from a Chinese college student's death after clicking on a fraudulent ad for cancer treatment. It is difficult to tell where ads end and legitimate search results begin, and many of the "legitimate search results" are owned by Baidu, under different brands.
In effect, Baidu is trying to get the user to click on a search ad. If that fails, get them to click on an "independent result" that is actually owned by Baidu, so that Baidu has another opportunity to serve ads (to be fair, Google also engages in the same practice, although perhaps less overt).
Unfortunately, questions about quality have persisted. In a recent post viewed more than a million times, a Chinese journalist posted a picture of a tombstone with Baidu on it, claiming that Baidu is dead because of its increasingly brazen commercialization practices and declining quality. This negative perception is further reinforced by the fact that many Chinese internet users welcome Google's return.
Ultimately, I have a hard time believing that an organization that beat Google 10-years ago, has 20-years of data, and that employs 4,500 AI researchers can't make a good search engine. The problem is likely a business decision (over-commercialization) rather than technical, and the pressure has likely been exacerbated by Baidu's slowing revenue growth. Perhaps this pressure will lessen if other business lines see increased success or the Chinese economy picks up, but ultimately it is up to the CEO / Founder, Robin Li, to enforce the culture he has already laid out: "If we lose our users’ support or fail to stick to our values, Baidu will be only 30 days away from bankruptcy." He knows what is at stake.
General Macroeconomic Risk
Speaking more broadly on China, the general slowdown of the economy, unsustainable debt loads, trade war uncertainty, aging demographics, and crackdowns on private businesses are all well documented macroeconomic factors and risks, but impossible for me to quantify.
Long term, however, I think it's safe to assume that the trajectory of the Chinese economy will be positive, the economic weight of the world will continue shifting to the East, online penetration will increase, some form of digital search will continue to be needed, China will continue to champion its tech giants, and Baidu will continue to be exposed to favorable emerging industries and opportunities.
Video / iQIYI
Revenue is also growing at 52% YoY, driven by a combination of paid subscription services, advertising, and licensing. However, costs are also rapidly increasing, as it continues to invest money into original content production and licensing, and it is still losing large amounts of money (-$1.3B in 2018).
It is not hard to imagine a path to profitability, however, and the rough business model has already been demonstrated by Netflix. Netflix has been able to raise subscription fees 10% per year for the last few years, while also growing paid memberships at 25%:
Netflix earns ~$113 average revenue per user (ARPU), compared to iQIYI's $41 ARPU. The revenue models are not directly comparable because iQIYI relies more heavily on advertising and operates in a less developed and more competitive market, but even so there is likely plenty of room to grow ARPU over the long term.
Netflix is targeting 13% operating margin in 2019. Long term, being fairly conservative, I do not think it is unrealistic to assume that iQIYI will reach at least 5% operating margin, which is what I assume in my valuation below.
Artificial Intelligence as a Service (AIaaS)
Baidu is heavily promoting itself as an AI first company, and is staking its future on its ability to offer AI powered services, primarily for enterprise customers. I've written in the past about how we have to be careful with AI hype, but I believe that Baidu's strategy is a sound one, and they are already generating revenue (~$500MM in 2018, growing 100% YoY).
In my humble opinion, the near term, commercial future of AI won't come close to the commonly envisioned Artificial General Intelligence popularly associated with Skynet. Intelligent robots won't be taking over the world anytime soon. Instead, AI will be used to solve relatively mundane (but profitable), day-to-day business tasks: automated speech-to-text translation, language translation, face recognition, etc.
To get a better idea of the breadth of potential applications, some of the AI products Baidu offers include:
- Drivers license identification
- Vehicle identification
- Porn identification
- Animal identification
- Language translation
- Insurance risk pricing
- Fraud detection
- Speech recognition
- ...and dozens more
These technologies will be consumed as a service - i.e, most companies will not develop them in house, but simply send their data to Baidu (and other cloud companies) to be processed and analyzed in near real-time. Baidu will make money by charging to process the data, and because AI services are higher up the value chain than typical "first generation" cloud services (like commodity compute hosting), AIaaS is likely to enjoy higher long-term margins.
To be conservative, however, I assumed a long term operating margin of 20% for Baidu's cloud AI division (even though Amazon Web Services currently has a 29% operating margin).
Baidu also has a few other "moonshot" opportunities:
- DuerOS - A voice assistant / speech recognition technology (think Alexa) with over 204MM users.
- Apollo - An autonomous driving platform with 135+ industry partners, including Volvo and VW.
- Ctrip - A 20% stake in the leading online travel agent in China.
All of these opportunities provide strong optionality, but short term I've assigned no value to DuerOS, a token $1B valuation to Apollo (based on similar private valuations), and a greatly reduced valuation to Ctrip (which I believe is overvalued by 2x+, even though it is publicly traded).
Long term, I believe most of the value in these opportunities will be created by the data. Through it's various business lines, Baidu will have access to large repositories of: voice data (DuerOS), driving data (Apollo), video data (iQIYI), and search data (Baidu search).
This data will be used to continually train their AI services, making them increasingly better. As the services improve, more people will use them, giving Baidu even more data to crunch. This is a positive, virtuous cycle that will generate a wide competitive moat that cannot be overemphasized - even though little value is being assigned to it now.
My valuation utilizes a discounted cash flow model and can be viewed here:
The first sheet contains my growth assumptions, the 2nd sheet contains my valuation inputs, and the 3rd sheet contains the valuation output. Feel free to copy the spreadsheet and play around with the numbers if you don't agree with my assumptions.
These spreadsheets are roughly based on Aswath Damadoran's publicly available valuation models. For a good overview of his valuation philosophy, see his classes on valuation here.
The main modification I make is to eliminate his use of beta (and the broader financial industry's use of beta) as a way to measure relative risk. Instead, I use a neutral beta of 1.0 for each company, which does not impact my discount rate and which allows me to compare valuations, apples-to-apples, across my entire opportunity set. Intuitively, this also makes sense because I view the discount rate as my personal "opportunity cost" of not investing in the broader market (i.e. an index fund).
To account for relative risk, instead I look at the valuation of a stock relative to its price. For "riskier" stocks, I demand a greater "margin of safety." These definitions are pretty subjective, relative, and loose...but I'd have a hard time trying to quantify the risks above (or the risks of any company) in an absolute, precise number.
Legendary value investor Howard Marks says:
If I could only ask one question regarding each investment I had under consideration, it would be simple: How much optimism is factored into the price? A high level of optimism is likely to mean the favorable possible developments have been priced in; the price is high relative to intrinsic value; and there's little margin for error in case of disappointment. Buf if optimism is low or absent, it's likely that the price is low; expectations are modest; negative surprises are unlikely; and the slighitest turn for the better would result in appreciation.
I believe my growth assumptions are realistic and conservative. I'm assigning a fairly pessimistic long term growth rate for the core search business, average prospects for the growing video and cloud businesses, and almost no value to the moonshots. Even with these assumptions, the stock seems 20% undervalued.
It feels like there is little to no optimism baked into the price, and "that the slightest turn for the better would result in appreciation." There will likely continue to be short term volatility, but long term the risk / reward trade-off seems sensible.
Thanks to David (my brother) for the tip on this stock!
Disclosure: I am long BIDU.