Description

   There are no assurances that securities identified will be profitable investments. The securities described are neither a recommendation nor a solicitation.

Company Description

Baidu is best known as The Google of China, however, it also has leading positions in several other important parts of technology as well as a large portfolio of net cash and investments. It is mispriced because it trades on EPS when it should trade on a combination of EPS for its search business and asset value for its long-term investments, “other bets”, and net cash. Net cash and investments comprise 81% of its market cap before even considering the value of their other bets. Backing out cash and investments, their core search business trades for just 3x 2018 EPS which includes heavy discretionary spending on “other bets”. Backing out our estimate of value for their “other bets” and you’re getting The Google of China for free.

Baidu has two operating segments, Baidu Core and iQiyi. Baidu Core is comprised of:

Baidu search & feed – 75% market share of search in China.

Baidu maps – The Google Maps of China. 70% market share.

Apollo – The Waymo of China. China’s national champion in autonomous vehicle technology.

DuerOS – The Alexa of China. China’s leader in voice recognition and smart speakers.

Baidu Cloud – The Google Cloud of China. Cloud services (a small AWS).

Haokan and Quanmin– short video apps. Baidu’s response to TikTok.

The second segment is its 57% stake in iQiyi, “the Netflix of China,” with 100MM paid subscribers (for reference, Netflix has 152MM), which is consolidated in Baidu’s financials. Baidu’s stake has a public market value of $19 per share; however, because it is consolidated with Baidu it does not appear on the balance sheet as a long-term investment, and because it is losing money, it is a significant drag on Baidu’s reported EPS.

Baidu has a large number of non-earning assets including a 19% stake in Ctrip (the Bookings.com of China), a small stake in Didi (the Uber of China), China Unicom, NetEase Music, and about 74 others. These investments total $34 per share. Baidu also has net cash of $11B, or $30 per share. (post script -Baidu sold their stake down to 13% as we were going to press, for $1B in cash).

As a result of consolidating iQiyi and holding a significant number of non-earning assets, Baidu is easily misunderstood. It doesn’t screen well on earnings metrics. As you might guess, analyzing Baidu from a corporate level is messy due to the corporate structure and stream of corporate activities. Financial statement issues that make the analysis tricky include net income attributable to non-controlling interests in consolidated entities (iQiyi), equity method accounting (Ctrip), consolidating the losses of “other bets” in Baidu Core, non-recurring gains/losses from divestitures (Du Xaoimin, Qunar, Xaiodu, iQiyi, others), and non-earning long-term investments and net cash that equate to 81% of the market cap. These issues distort Baidu’s earning power and make it difficult to estimate the company’s intrinsic value. We untangle this ungodly mess by analyzing each piece separately, then value the company as a sum-of-the-parts.

Investment Thesis

We’ve grouped our analysis into five parts which will comprise our sum-of-the parts:

1. Search ecosystem (including feed, maps, short video, and mini programs)
2. Cash & Investments (iQiyi, Ctrip, others)
3. Apollo
4. DuerOS
5. Cloud

1. Search

The online population in China is enormous - 829MM out of a total population of 1.38B, resulting in approximately 60% penetration, in contrast to 90% penetration in the U.S. Rising penetration provides a modest tailwind. The online ad market is also huge yet has continued to experience rapid growth in recent years (approximately 30% growth in 2017, 31% in 2018, and 28% in 2019e). China digital ad spending as a percentage of total ad spending has grown from 30% in 2013 to approximately 68% today. As a result, digital ad spending growth must slow soon to a rate that more closely reflects total ad growth.

Source: Bernstein

Similar to Google, Baidu generates most of its revenue from online marketing services via its search engine, mostly auction-based pay-for-performance services, enabling customers to bid for key words and priority placement of their links in Baidu’s search results. Typically, customers only pay Baidu when users click on one of their links (on Baidu’s search results page or an advertising partners’ page). Online advertisers are focused on the ROI on their marketing dollars and go where they get the biggest payback, which typically is a function of the number of the users and the quality of the lead (relevancy). Search generates high ROI’s for advertisers because it provides a large audience with highly relevant ad placement. Baidu has the largest user database in China which it has built over the last 19 years. This allows them to make more personalized (relevant) recommendations tailored to the user’s interests, in turn generating a better user experience and a higher ROI for advertisers. Advertisers are often willing to pay higher prices for Baidu search ads because of the large audience they provide and the high relevance and targeted nature of the ad and its placement. In summary, search is a proven and effective method of advertising and Baidu is the dominant provider of search advertising in China.

Baidu has a huge competitive advantage - the bigger the user database, the better the algorithm, the better the search results - which is a network effect difficult for competitors to breach. Baidu has over a billion consumers in Asia feeding its AI-driven search algorithm, allowing it to continuously improve search results. Better search results beget more users which beget better search results...the network effect. Further, Baidu invests heavily in R&D at 17% of sales and has over 4,000 patents, which only increases the barriers to entry.

While Baidu has a near monopoly on internet search in China, its position is still probably not as strong as Google’s is in the US. As VIC member mackam836 noted in his/her June report on Baidu, Alibaba blocks Baidu’s web crawlers from indexing its pages for product searches. As a result, most product searches in China are conducted on Alibaba or JD.com. Further, China’s mobile market is moving away from apps and towards a few super apps such as WeChat and Alibaba, which are creating content ecosystems via mini-programs. More on this later, but in some cases this may lessen the need to leave the super app to do a search.

Ok, enough background, onto the details. Baidu search accounts for most of Baidu Core segment revenues and well over 100% of its earnings. Unfortunately, within the Baidu Core segment the company doesn’t break out the profitability of search from the losses in their “other bets”, so we do not know the stand-alone profitability of any of the major businesses within Baidu Core – search, Apollo, DuerOS, and Baidu Cloud. So we start by referencing historical margins during periods when spending on “other bets” was lower, as well as taking a look at comps. In general, search is thought of as a 50% margin business. Google’s core search margin (also obscured by their other bets) is widely considered to be >50%. Russian search engine Yandex has consistently generated operating margins in search of 30-35%. Baidu’s historical margins are consistent with this with ~70% gross margins and ~50% operating margins during normal times; however, the company has a history of making “other bets” that periodically depress Baidu Core’s high margins. Take a look:

  Source: Company filings

As you can see, the Baidu Core segment has historically had GAAP operating margins of roughly 50% in years without extraneous spending on other bets and roughly 30% margins in years when extraneous spending occurred. In 2019, Baidu has experienced extraordinarily weak profitability with margins falling into the mid-teens. This has been driven primarily by weak monetization of Baidu search. A large gap has recently developed between continued healthy user statistics and the ability to monetize that traffic, i.e. sales growth has slowed and margins have declined meaningfully this year, despite continued growth of its user base.

Baidu’s user engagement appears healthy, with key metrics such as Daily Average Users (DAU) and number of search queries maintaining healthy growth. Specifically, the Baidu app’s DAU increased 27% in 2Q. Search queries grew 20%, an encouraging sign their dominant position in core search is intact. The problem has been monetizing this traffic which the company attributes to:

• Greater competition - a huge surge in ad inventory, presumably from rapidly-growing TikTok, amidst a weak ad market. This has led management to increase spending on traffic acquisition costs (TAC), feed, mini-programs, and short video, as part of their recently announced “growth over margins” strategy.
  
• A weakening macro and softening ad market.
• Self-inflicted disruption caused by a proactive shift to structured data landing pages for healthcare queries (which accounts for 1/7th of search revenue) made to improve regulatory compliance, user experience, and advertiser ROI.
• Recent government policy changes in online games and financial services verticals.

Let’s address each of these:

Greater competition

The biggest competitive impact Baidu has seen recently has come from the growth in advertising on entertainment and social media platforms, particularly TikTok (a very popular short form video platform similar to Vine in the US, owned by ByteDance). As TikTok and other news, entertainment, and social apps and feeds have gained in popularity, search has lost share within online advertising. This increased competition within online advertising and increased digital ad supply in China has hit Baidu particularly hard in 2019.

Search used to be the go-to source of information on the internet; however, the emergence of popular feeds are resulting in more "push" and less “pull” of online information. According to iResearch, the market share of search ads in the China internet ad market has declined from 33% in 2014 to 20% in 2018, and is expected to continue to decline, as news and entertainment feeds have exploded in popularity and taken massive share (see below).

           Source: Goldman Sachs

However, while internet users are expected to continue to shift their time to feeds, the search ad market is still expected to grow in absolute dollars despite losing share of the overall online ad market, providing Baidu a backdrop for continued growth.

Source: Oppenheimer

To wit, the Baidu search ecosystem continues to exhibit good growth with search queries up >20% in 2Q, an increase from mid-teens growth in Q1. Monthly average users (MAU) were up 18%, continuing a trend of mid- teens organic growth.

Source: Citi

As we can see, Baidu’s recent weakness in online advertising revenue and margins does not stem from user activity. It is not losing its share of search, only its share of ad dollars. In 2Q19, Baidu reported that while the overall online advertising industry market grew between 10-20%, ad inventory more than doubled year-year which severely impacted CPMs (pricing). The flood of ad inventory looks to stem primarily from TikTok, which more than tripled its revenue last year, grew 129% in 2Q19, and is targeting to double it this year, taking massive share of the online ad market virtually overnight. Its share of all online advertising dollars has gone from <1% in 2015 to around 20% today.

            Source: Bernstein

The weakness seems to have peaked in 1Q19, when Baidu reported weak results citing a “flush of new ad inventory” from social media platforms. As seen below, incremental ad supply coming primarily from the growth in feed products has increased competition within the online ad industry.

      Source: Goldman Sachs

With TikTok (Bytedance) increasingly capturing user time, and Tencent ramping up monetization, Baidu is losing share. In response, Baidu has shifted its near-term strategy to “growth over margins” via increased investment in feed, short video, and mini programs, to increase traffic and user engagement in an effort to recover market share of online ad spending. This spending has depressed profitability in the near-term. It is difficult to determine whether this strategy will succeed. They are late to the game in feed and short video, and the odds are stacked against them. However, if the strategy fails we believe management has the discipline to halt spending on these initiatives and return to focusing on core search. They have shown that type of discipline on many occasions in the past. A sole focus on core search would provide a meaningful lift to profitability if the spending on these initiatives are halted. In other words, even if the current level of competition persists, margins should normalize to over 30% as they lower investment spending and/or increase monetization.

In addition to suffering from greater competition from TikTok, the China internet landscape is morphing where a few large players like WeChat and Alibaba are becoming “super apps”. A super app is somewhat unique to China and is really just developing now. A super app is essentially one big umbrella app that houses many small apps, called mini programs, which allow users to navigate within one super app instead of leaving the super app portal, in turn allowing users to get rid of memory eating apps on their mobile devices that they may only use a couple times a year. They are a one-stop online hub that allows users to bypass other websites and disparate interfaces. Rather than having to download an app, mini programs are all contained within the one seamless interface of the super-app. For advertisers and merchants, mini programs lower the customer acquisition cost and improve conversion by allowing purchase without leaving the super app.

The advent of mini programs enabled the development of super apps. WeChat is arguably the world’s first super app. In 2017, WeChat launched the mini program which allowed developers to develop apps within the WeChat ecosystem. Mini programs (or mini apps) are less than 10MB, can run instantly, and can take advantage of capabilities already within the super app ecosystem, like payments capabilities. For example, once you have registered a credit card with WeChat, you can use all of its payment capabilities to, for example, book a flight, order a cab, order takeout, pay a bill, or make a dentist appointment. A WeChat in-app shopping marketplace via JD is already one of the largest ecommerce platforms in China. In just two years, the number of WeChat mini programs has grown to 2.3MM, more than the number of apps available in Apple’s app store. Many technologists believe super apps with mini programs will define the future of the internet in China.

Baidu was late to the mini program game and lacks the first mover advantage of WeChat, but it is quickly catching up. In 2Q19 Baidu’s mini program MAUs grew to 270MM. Monetization should just be a matter of time. Baidu mini apps could provide a unique value proposition for advertisers and merchants because WeChat is mainly social-oriented, whereas search is intent-based.

Baidu was also late to launch a news and entertainment feed and is now playing catch-up. Baidu’s feed revenue is expected to grow approximately 50% in 2019 after growing 106% in 2018. Adding feed to search increases time spent on the Baidu app significantly. Management believes they are succeeding in feed, but as noted earlier, if they don’t succeed the investment in feed can stop, which would likely provide a nice lift to margins. Their investment in feed is large and discretionary. Bernstein, which thinks Baidu is too late to the game in feed and should abandon it and focus on search, estimates if Baidu shut down feed they could return Baidu Core to a low 30%’s operating margin (equivalent to EPS of $10.00-11.00 just for Baidu Core and including all spending in “other bets” like Apollo and DuerOS).

Short video is yet another area in which Baidu is playing catch-up. According to QuestMobile, Baidu’s two short video apps (Haokan and Quanmin) have MAU of 85MM. For perspective, TikTok parent ByteDance has 1.5B! Tencent’s short video app has 105MM MAU. Baidu is playing from way behind in short video and likely losing a lot of money in this area. This is of course another source of upside if they simply shut down their short video efforts. To reiterate, management has demonstrated that if initiatives like these do not work, they are not shy about shutting them down or selling them. In fact, Baidu already scaled back its marketing spending in 2Q19 in areas that did not meet their ROI criteria.

In summary, Baidu is shifting from being a search-only provider to a super app with search + feed + mini- programs + short video. Its AI technology combined with their huge search database should allow them to catch up to the competition in some areas and narrow the gap between user experience and monetization. While monetization has definitely weakened and competition has increased, user engagement is still very high, indicating a healthy ecosystem and a place in China as one of only a few super apps. If competition persists and they cannot make headway in their efforts to improve growth in online advertising, they can reduce investment spending which should allow them to bring margins back to a more normal level.

Macro slowdown

If you buy into the Kyle Bass narrative on China macro, this idea is probably not for you. We have no great insight into this part of the discussion. China clearly has increased leverage which may result in a slowdown and/or credit issues at some point. Regardless, internet activity should grow over the next decade under most plausible scenarios. That said, the online ad market has definitely slowed this year due to both weak demand resulting from macro uncertainty and excess ad supply from growing media platforms. This was witnessed across many ad market participants in 2Q, a few of which are listed below along with their 2Q19 ad revenue growth rate:

Source: Company filings

Importantly, these competitor results lend validation to the argument that much of Baidu's weakness stems from a weak ad market and excess ad inventory - industry issues rather than a company-specific problem.

Healthcare Initiatives

Baidu’s reputation was damaged in 2016 when Wei Zexi, a 21 year old student, died after receiving an experimental cancer treatment he learned about from a promoted search result on Baidu. Unlike Google, Baidu’s promoted search results were not clearly distinguished from other results. State TV and the public broadly criticized Baidu’s for-profit business model and its lack of control over medical and healthcare-related advertising. Following the incident, regulators ordered Baidu to attach clear markers on ads, reduce the amount of promoted results to a maximum of 30% of the page, and ban ads from unqualified healthcare services. The new regulations came into effect on September 1, 2016, significantly impacting revenues from their healthcare vertical which accounted for 20% of search revenue at the time.

This episode has since faded considerably; however, Baidu continues to make modifications to improve its content, user experience, regulatory compliance, and advertiser ROI within the healthcare vertical. In March 2019, Baidu decided to proactively shift healthcare keyword queries to a new structured data landing page, allowing them to better manage and monitor healthcare-related content. The new landing page means Baidu users will stay on Baidu’s website or app rather than being redirected to an advertiser’s site. This is expected to improve user experience as well as monetization in the long-term, although it has temporarily resulted in weaker ad sales among healthcare customers who are now required to provide greater disclosures than before. While it’s painful near- term, management believes it ensures better quality content and improves user experience by enabling them to better compare and find relevant healthcare services and products. Baidu expects the transition to take ~1 year.

Regulatory

On the 2Q19 conference call, management shared that its top 12 ad verticals contribute about two thirds of total ad revenues. While half of them recorded very healthy growth in the mid-teens, the other six experienced some sector-specific headwinds. For example, healthcare is going through its structural data initiative, the gaming industry was forced to remove what the government deemed to be socially harmful or inappropriate content (e.g. violence), the auto is suffering from weak end markets, and the financial services industry encountered some new regulations. Excluding these four segments, growth would have been in the mid-teens, offering some validity to their comments that there are some temporary factors related to the economy and regulation impacting revenue growth this year.

That’s a long summary of Baidu’s core search business and the headwinds it faces. In short, they remain the only viable search alternative in China, but must continue to invest to remain competitive in an online ad market which is increasingly crowded.

So what’s it worth? Estimating the value of Baidu search requires the investor to make some adjustment to the Baidu Core segment. Specifically, 1) normalize margins to reflect long-term spending and monetization levels for the search ecosystem (feed, short video, mini programs) and 2) back out (undisclosed) investment spending on “other bets” (Apollo, DuerOS, Cloud). These are difficult to estimate, so we go back to referencing historical Baidu Core margins of 50% during periods when they lacked unusual investment spending. Given greater competition today and the risks associated with its ever-changing nature, we believe normalized search margins are likely to be lower than historical levels, but are unsure of the degree. We know for sure there is a lot of discretionary spending today on their growth initiatives previously discussed, we just don’t know how much. We also don’t know if their initiatives will succeed, but importantly, we don’t have to because 1) core search still appears healthy, 2) management is rational and will likely not pursue opportunities that aren’t meeting their ROI requirements and 3) the margin of safety in the stock price is large enough to allow for significant error. Below, we show a range of values for Baidu Core (excluding other bets) using a very wide range of possible operating margins and multiples. Anything below a 30% GAAP operating margin would imply a broken business model and/or prolonged discretionary investment spending in a failed strategy with no payback, which we believe is unrealistic.

The boxed-in values in the upper part of the table represent the approximate range of sell-side estimates while the lower box represents our estimates. We believe the sell-side uses a margin assumption of 15-30% based on current results and inclusive of all of the heavy discretionary investment spending in Apollo, DuerOS, TAC, feed, short video, and mini programs. It then assigns a 10-12x P/E despite Baidu being a highly profitable monopoly. This puts the sell-side’s estimate of Baidu Core at $49-$116 per share. Implicit in this is that the core search model is permanently impaired and that Apollo and DuerOS are worth less than nothing and are permanent investments in failed strategies. All of this of course flies in the face of historical precedent with management backing off investment spending on multiple investments, and with the significant value created investing in OTA and video via their multi-billion dollar equity stakes in Ctrip and Apollo. This also ignores any chance of success in normalizing search monetization. Excluding Apollo, DuerOS, and a temporary decline in monetization, and considering where there margins were just a year ago prior to an increase in spending, we believe normal GAAP operating margins are 35-45%. Historical margin levels and comps lend credence to this range. Greater competition and higher stock-based compensation is a 5%-15% margin headwind versus historical levels. At 17x, a more appropriate multiple for a dominant search engine in a large and growing economy, Baidu search is worth $220 per share. We are dealing with imperfect information and admittedly could be wrong, but as you will see the margin of safety in the stock price captures the inherent risk to our estimates.

Value for search $220 per share

2. Equity Investments

As of 2Q19 Baidu has $30 per share in net cash and $34 per share in long-term investments. These totals deconsolidate iQiyi and are for Baidu only.

The public market value of Baidu’s 57% stake in iQiyi is worth an additional $19 per share. In total, Baidu has $82 per share in net cash and investments. Their long-term investments include a 19% stake in Ctrip worth $11 per share and a small stake in Didi worth $4 per share. Other notable investment holdings include China Unicom, Du Xiaoman, Qeeka, Uxin, HAND, ele.me, Netease Music, and WM Motor. According to Crunchbase, they have 78 investment holdings in total. Some of these investments came about from combining their “other bets” with competitors and taking equity in the newly formed company, some are equity investments made as part of a partnership, and some are part of a push into AI through its $1.5B venture capital fund, The Apollo Fund. Almost all of these investments are ignored in sell-side valuation models.

It is difficult to know precisely how successful their prior investments have been without knowing their total investment; however, their monetizations appear to have generated returns within the realm of reasonable. Their online travel agency (OTA) business, Qunar, was combined with Ctrip and is now worth $3.9B. Their 57% stake in iQiyi, mostly self-funded, has a market value of $6.7B. Their ride sharing business was combined with leader Didi and their stake is now worth $1.4B based on the last funding round. Importantly, they have shown a willingness to opportunistically monetize businesses – food delivery, financial services, and other online-to- offline (O2O) services. Just a few weeks ago they monetized about a third of their stake in Ctrip for approximately $1B cash.

Many of their long-term investments have readily determinable values. Those that don’t are marked at cost. As such, we are accepting the stated values of their investments at face value. Further comfort comes from their history of responsible investing with reasonable outcomes.

Our estimate of the total value of their net cash and investments is:

3. Apollo

“If the driverless-car phenomenon takes off in China, the payoff could be in the trillions of dollars.” – McKinsey January, 2019

Highlights from the McKinsey report:

1. Sales of autonomous vehicles (AVs) in China to reach $900B by 2040.
2. An additional $1.1T opportunity from mobility services by 2040.
3. AVs will be 40% of unit sales and 12% of the car park by 2040.
4. Mass adoption of AVs will inflect around 2025-2030.
5. China will likely emerge as the world’s largest market for AVs.

China has a stated goal of becoming the world leader in Artificial Intelligence (AI) by 2030. China is investing tens of billions of dollars in AI, far more than any other country. Perversely, China may be able to make faster progress in AI than its U.S. peers due to a lack of regulation. They have the ability to test technologies in a real world environment without as many regulatory barriers as you would find in other countries. For example, in recent months the Chinese government has increasingly eased regulations on AVs in a number of cities to accelerate the path to commercialization. Within weeks of Uber’s self-driving car fatality in the US, China granted local governments the authority to allow AV testing on public roads. Decisions like these create a notable advantage for Chinese AV companies given that testing is only allowed for domestic vehicle manufacturers. Further, China’s state sponsorship and promotion of “national champions” creates a formidable barrier to outside competition and tends to lead to industry dominance for the designated champions. In November, 2018 the government designated Baidu as the “national champion” in autonomous driving. They continue to work together to bring AI to urban areas with plans for autonomous city buses and public digital city maps.

Baidu is the market leader in AI and AVs in China. Baidu already has more AI-related patents than any company in the world in areas such as natural language processing, smart search, speech recognition, and autonomous driving. Its ownership of huge proprietary data sets from two decades of accumulating search activity further provides a significant competitive advantage. Implementing machine learning in 2010 and deep learning in 2012 in their AI-powered search algorithm also provides them with first mover type benefits. Lastly, open-sourcing their AV software has allowed Baidu to create a vast web of partners and joint ventures, similar to what Google did with Android. Given all this, the company seems to have a shot at being a world leader in AI.

The two primary AI initiatives housed within the Baidu Core division are Apollo and DuerOS. Apollo is Baidu’s open source autonomous driving platform. Baidu has called the platform “the Android of the autonomous driving industry”. Apollo is primarily focused on software although they also develop hardware products such as cameras, sensors, and radar. The company has pointed out that Beijing drivers spend 30% of their drive-time looking for a parking space, just one small data point, but a good example of why AV software should find demand. They are currently working on this specific issue and hope to have a helpful solution soon.

As of July 2019, Baidu had received almost half of the 204 autonomous driving pilot licenses granted in China, 5x that of the next largest competitor. It is also the only company in China to receive T4 licenses, the highest level of autonomous driving test license issued by the government. Baidu recently revealed that it has tested 300 level four autonomous vehicles across 13 cities altogether, achieving 2MM km driving in urban-road environments. Driver data is critical for developing self-driving technology as data trains the algorithms that guide the autonomous vehicle. According to the company, Apollo has amassed over 10x the test miles of the next closest competitor.

Largely because Baidu open-sourced their platform, it now claim’s one of the world’s largest partner ecosystems for autonomous driving.

OEM partners include major manufacturers such as Toyota, BMW, Ford, Hyundai, and Volvo. Volvo will be its first global OEM partner to integrate Apollo software into their EVs in 2022 when they plan mass production of self-driving EVs in China. Baidu plans mass production of a level four autonomous commercial EV with Hongqi (luxury car) in 2020.

Apollo’s Apolong minibus, a level four autonomous vehicle, has been operating for over a year with zero accidents and over 300,000 people served in 10 locations (airports, etc.). Although clearly it merits no style points.

So what is Apollo worth? This is a difficult question. It’s first important to note that investors currently ascribe a negative value to Apollo because it generates losses and is part of Baidu Core, which investors generally value as one segment. As noted earlier, Baidu doesn’t break out search profitability from the losses in their “other bets” within the Baidu Core segment, so we do not know the stand-alone profitability of any of the businesses within Baidu Core, including Apollo.

Here is what we do know: 1) there are little or no revenues today, 2) there is heavy investment spending of an unknown amount, 3) future profitability is probably a long way out and highly uncertain, 4) there is potentially a huge TAM, and 5) there is a lot of hype.

Comps are worth considering.

Waymo (Google) seems to have a first mover advantage in the US - it has been around for the longest time and has by far the most data and miles driven. Waymo booked 1.2MM miles driven in California last year (where most AV testing is done). Waymo is ahead of competitors, but China is always going to favor the home team so there is room for Apollo to succeed in China. In March it was rumored Alphabet was seeking outside capital for Waymo, but funding was not secured so we still don’t have our first mark. Sell-side estimates of value for Waymo range wildly from $0 to $250B. Just a few weeks ago Morgan Stanley lowered their estimate of value for Waymo by 40% to $105B using a ridesharing-driven DCF!

Cruise was acquired by GM in 2016 for $1B. In May, 2019 Cruise raised $1.2B at a $19B valuation from a group of credible investors (e.g. T. Rowe, Honda, and parent GM). Miles driven in California were 450,000 last year. It has 1,500 employees versus 40 at the time of acquisition in 2016. It has an annual budget of approximately $1.2B, possibly a good reference for Apollo’s level of spending. It is probably the best comp for Apollo given its size and stage of development, although there are admittedly big differences between them such as the size of the China market, Apollo’s leadership position in China, etc.

Argo is Ford’s AV unit and has been valued at $7B via a recent investment by VW and Ford. These OEMs are investing heavily behind Argo with a big recent investment, but it seems far behind the competition today.

Goldman values Apollo in their SOTP model at $3.5B, or $10 per share, assuming PV of 30x their 2021 sales estimate. That doesn’t make a ton of sense to us given the sales and earnings lay far beyond 2021. Bernstein takes a stab and comes up with an $8.8B valuation, or $25 per share, assuming a PV of 25x Apollo 2033e earnings. Their approach is built from the bottom-up and seems rational. We adopt their framework for valuation and plug in our own numbers (below). Other sell-side analysts ignore it and implicitly value it at a negative value by including the investments in Apollo (losses) in Baidu Core operating earnings.

Industry consultants expect sales of autonomous vehicles to reach approximately 5MM by 2030. For perspective, auto sales in China are approximately 29MM annually today. This would equate to 19% penetration in 2030 (assuming 4% growth in the car parc). At an average cost of $12,000 for an AV (they have very small cars), the revenue opportunity for AVs would be $60B in 2030. Baidu expects Apollo to get 15-20% of the MSRP, for a revenue opportunity of $9B to $12B. A 50% market share reduces the opportunity to $4.5B to $6B. With a dominant position and high margin software sales, Apollo should be able to generate an operating margin north of 30%. Assuming a 30% margin and 17% tax rate, they could generate net income of $1.1B to $1.5B, or $3.13 to $4.26 in EPS. At 20x, reasonable for a dominant software company, Apollo would be worth $63 to $85 per share in 2030. Discounted back to today’s dollars at 10% we arrive at a range value for Apollo of $24-$33. There are a lot of difficult assumptions to make and we try to avoid false precision, but this may be a framework to use as the business develops.

Our final estimate of value for Apollo uses an average of the Cruise valuation, Bernstein’s estimate, and our own bottom-up analysis. We then haircut the valuation by 30% to reflect the hype and extremely high level of uncertainty. Totally arbitrary we know.

Apollo at Cruise valuation $19.0B

Bernstein $8.8B

Bottom-up valuation $10.0B

Average $12.6B

“risked” at 30% -$3.8B

Apollo value $8.8B

Per share $25.00

Value of Apollo $25.00

4. DuerOS

DuerOS is Baidu’s voice assistant and smart speaker, similar to Amazon’s Alexa/Echo. It has a large and rapidly growing installed base of 400MM devices, up 450% from last year, and 3.6B monthly voice queries, up 750%. Among smart speaker manufacturers, DuerOS ranks first in market share in China and third worldwide. It has garnered 16% global market share despite launching less than two years ago. It executed on the smart speaker land grab incredibly well.

     Source: Citi

DuerOS plays a key role in Baidu’s strategy to expand Baidu search beyond mobile and into home appliances and car infotainment systems. The low cost of smart speaker devices allows Baidu to become the point of access to the internet as search goes from mobile to voice. If Baidu locks up the home and car with DuerOS it could become a significant franchise with large barriers to entry. DuerOS is building a lot of AI into DuerOS, thus the name “smart speaker”. It can recognize who you are, identify a child’s voice so that it only recommends appropriate content, and needs no wake up word like “Alexa” or “Siri”. Baidu has partnered with large hotel chains and auto OEMs to embed voice assistants in hotel rooms and car’s infotainment systems.

The company expects heavy investment spending to continue this year to further grow the installed base. Similar to Apollo, Baidu has been investing heavily in DuerOS although the level of investment is unknown because the business is consolidated with the Baidu Core segment. Also similar to Apollo, value may not be recognized by the market until a monetization event. At present, DuerOS is only monetizing via low-margin hardware, but expects “meaningful monetization” to ramp up next year. The primary method of monetization will be advertising and revenue sharing from music and audio book apps.

DuerOS is another business that’s very difficult to value because we don’t have independent financials. It’s consolidated into Baidu Core. There are no sell-side estimates and similar to Apollo, they implicitly value DuerOS at a negative value by putting a multiple on Baidu Core when DuerOS is losing money while in investment mode. We believe DuerOS has value but we don’t have any basis for valuation - no comps, no financials, not even a revenue number. It wouldn’t move the needle too much on the intrinsic value of Baidu, and it wouldn’t change our conclusion, so we will value it at $0 for now.

DuerOS value $0

5. Baidu Cloud

Baidu Cloud is another young “bet” within the Baidu Core segment, currently annualizing at an $888MM revenue run rate and growing 90%. Similar to feed, short videos, and mini-programs, they are playing from behind but are doing so to stay competitive. Their strategy is to offer enterprise AI solutions as a key competitive differentiator. The few sell-side estimates on Cloud value it at 2-4x 2020e sales, or $9-18 per share. Others value it as part of Baidu Core, implicitly resulting in a negative value for this division as it spends heavily during its growth phase. Until profitability metrics are better known, we’ll value Baidu Cloud at 1x sales, or $888MM.

Value of Cloud $3

 Sum-of-the-parts

Our sum-of-the-parts results in a total estimated value of:

Baidu is cheap, very cheap, which provides us with a large margin of safety. Its $108 stock price shrinks to only $26 when excluding its net cash and investments. It’s hard to find the downside short of a sovereign crisis or fraud.

The biggest controversy is clearly in the value of search. The value of search ranges from $30/share if permanently impaired by industry disruption, to >$300/share if they return to 50% margins and a respectable valuation deserving of a business with huge network effects and monopoly-like characteristics. If we use sell-side estimates for search of approximately $71 (11x EPS using only a 20% margin) rather than our estimate of $222 (17x using a 40% margin), then add the other assets and bets, we still get an intrinsic value of $181 for Baidu, 76% upside.

Why does this opportunity exist?

We think the biggest reason investors have a hard time valuing Baidu is because the valuation analysis is part asset value, part earnings stream (which are depressed). Book value is a meaningless concept to most tech investors and growth investors, but it shouldn’t be ignored here because it’s a big part of the thesis. Baidu has witnessed dramatic growth in book value over the last several years reflecting an accumulation of cash and investments, yet the stock has declined over 30% during this period.

   Source: Factset

Now reference the charts below. The first chart is the NTM P/E ratio of the three public search companies - Google, Yandex, and Baidu. As you can see, they tend to loosely trade together and, over time, at similar average multiples. This tells us investors are valuing them on EPS and possibly ignoring meaningful asset value that Baidu has accumulated on the balance sheet as well as in “other bets’ that aren’t yet showing up on the balance sheet. Now reference the second slide, which is the same three companies on trailing P/TBV. As you can see, Baidu has completely de-rated on tangible book value, now trading at only 1.4x. This is further evidence that investors are only focused on earnings and are ignoring significant asset value.  

 Source: Factset

 Source: Factset

Keep in mind that Iqiyi is consolidated and is a significant drag on Baidu’s EPS and book value.  If Baidu were to reduce its stake in Iqiyi from 57% to 49% it would be deconsolidated and would add $2.46 to Baidu’s EPS and roughly $18 to book value, bringing P/TBV down to only 1.1x.

Risks

Competition - Super apps, specifically WeChat and its mini-programs, may lessen the use case for search. News and entertainment are being pushed through feeds and video apps like TikTok, further weakening the use case for some search activity. Their strategy around feed, short video, and mini programs could fail and/or result in a sustained period of wasteful spending.

Brand - Baidu lacks a great brand in China. The 2016 Wei Zexi incident hurt their reputation. Further, some users don’t like the quality of Baidu search results, citing poor and biased results that promote its own products and services instead of more relevant results. Baidu is not well liked by some users and seen as inferior to what Google could offer if they came to market.

Macro - China is highly levered and could see a significant economic slowdown at some point.

It’s China! - We have gotten pushback on taking on any exposure to China given their hybrid Communist- Capitalist economy, hybrid open-closed economy, social issues and lack of freedoms, highly censored internet, etc. No matter what the value, there are going to be “never China-ers”…those that won’t buy Chinese equities at any price. While these issues do pose hurdles to the investment thesis, we would point out that their system (government sponsorship, foreign barriers) tends to create monopolies with virtually no competition. Further, China Inc. continues to court foreign investment which will only occur if their equities appreciate. Baidu employees get paid heavily in stock (which we account for), so their interests are aligned with ours.

The Great Firewall of China - Common websites we take for granted and use every day, like Google and Facebook, are banned in China because they are not censored. If you try to use them in China you will be blocked by The Great Firewall of China. Censorship is certainly a hot issue to debate in the U.S., but is part of China’s long-standing system and culture. Also, just look at what Facebook is going through today regarding censorship. This is a difficult, very personal issue. It’s all a matter of where you draw the line. There are degrees of freedom, even in the U.S. You can smoke pot but you can’t snort cocaine. Line drawn. You can drive 55 but not 80. Line drawn.

Google - Google showed interest in re-entering China a couple years ago with a censored search engine called Dragonfly, but any serious attempt was quickly dismissed when employees and others protested against the idea of having a censored search engine. Since then, trade tensions with China have risen. Taking this into consideration, it seems highly unlikely Google will re-enter the Chinese market anytime soon, if ever.

Corporate Governance – U.S.-listed Chinese equites became the poster child for public company fraud over the last decade. Transparency has also been an issue. It has improved in recent years, and Baidu is a large and well- established company, but the stigma remains and it says something culturally about corporate ethics in China. In order to be listed in the U.S., they do have to conform to our accounting standards and register with the SEC.

No voting rights – ADS holders have no voting rights. The security is technically a VIE, not a common stock.

Losing Share of Time Spent – Baidu has been losing share of time spent; however, we do not feel this is a relevant KPI. As users spend more time on the internet, they will likely watch more video and other media which uses large chunks of time. A search query requires very little time but remains highly relevant.

Sum-of-the-parts may not get recognized – In our experience, companies have a hard time achieving full recognition of the underlying value of their businesses when they are held in a conglomerate structure. The conglomerate discount may persist indefinitely and it may never trade at fair value, regardless of its intrinsic value.

Stock-based compensation (SBC) has risen a lot – SBC has risen from a 1% margin headwind five years ago to a 5% headwind today. So 50% historical normalized margins translate into roughly 45% margins today. The trend in SBC is disappointing, but we have accounted for it and put it in perspective of the total opportunity at hand.

Declining Smart Phone Sales – Baidu’s mobile app is typically preloaded on smart phones, so declining smart phone sales in China are viewed as an impediment to growth.

DISCLAIMER: THIS IS NOT A RECOMMENDATION. The securities described are neither a recommendation nor a solicitation. There are no assurances that securities identified in this note will be profitable investments. The stated opinions are for general information only and not meant to be predictions or an offer of individual or personalized investment advice. They are not intended as an offer or solicitation with respect to the purchase or sale of any security. This information and these opinions are subject to change without notice. Security information is being obtained from resources I believe to be accurate, but no warrant is made as to the accuracy or completeness of the information. Any type of investing involves risk and there are no guarantees. The author and its employer may or may not have material positions in the securities mentioned in this note and will not be obligated to give notice of any changes in their views or positioning. You agree to hold the author harmless and to waive the right to any legal action against the author in relation to this note. The views expressed here reflect the author’s personal views about the issuers and securities and do not constitute investment advice. The author makes no representation or warranty, express or implied, regarding the accuracy, completeness or adequacy of the information. Before making any investment decision, you should seek independent investment, legal, tax, accounting or other professional advice as appropriate, none of which is offered to you by the author in this note. The author accepts no duty of care to you in relation to investments. Past performance cannot be relied on as a guide to future performance. Any investment or investment strategy can be impacted by numerous factors, including market and economic conditions, and may result in a loss to investors.

Catalyst

*Recovery of search margin.

*Monetization events for Apollo, DuerOS, or investmnet holdings.

*Deconsolidation of Iqiyi.