November 2022 - AI Superpowers by Kai Fu Lee

This month we dive into head of Sinovation Ventures, Kai Fu Lee’s book on the future of AI. I disagree with a lot of this book, and overall found it underwhelming. However, there are some interesting ideas that people can take with them into the future.

Tech Themes

Competitive Intensity in China. China's tech industry is fiercely competitive due to the sheer size of its market and the government's support for innovation. Local players like Baidu, Alibaba, and Tencent (BAT) have dominated the industry for years, but new players are emerging every day. This intense competition has created a dynamic tech ecosystem where companies constantly innovate and disrupt traditional industries. The race to dominate emerging technologies like AI, cloud computing, and 5G is particularly intense as these technologies have the potential to reshape entire industries. The markets are so competitive that entrepreneurs use almost absurd tactics to beat out rivals. In one instance, a new social network named Kaixin001 was gaining in popularity. The company was new and didn’t have enough cash to buy the Kaixin.com domain name, so its number one competitor Renren built an exact copy of Kaixin001’s website and bought the Kaixin.com domain name to launch the product. Within months their traffic plummeted and although they eventually won a lawsuit for unfair competition, the $60,000 reward was nothing compared to the loss of customers. Renren itself was a clone of Twitter, started by Wang Xing, the eventual founder of Meituan. In another instance, Tencent and Qihoo 360 got in a repeated blame-game fight, that eventually led to Qihoo Anti-Virus blocking the use of Tencent’s QQ and Tencent suing Qihoo in the first-ever Anti-Monopoly court case. The Groupon Clone Wars were a series of intense price wars between Chinese group buying sites like Meituan and Dianping. At one point, China had 6,000 group-buying sites. These sites copied Groupon's business model of offering discounts on local goods and services but adapted it for the Chinese market. The result was a hyper-competitive market where companies would aggressively discount their services to attract customers. This competition was good for consumers but ultimately unsustainable for the companies involved. Meituan and Dianping have merged to form a dominant force in China's online-to-offline (O2O) market. This merger directly opposed Alibaba’s wishes, and it massively funded competitor Ele.Me to compete. Today, Bytedance and Alibaba are suing Tencent and Meituan over monopolizing the food delivery industry. The competition in China is absolutely brutal - copying isn’t just allowed, it’s encouraged.
AI in Practice. Meituan Dianping and Bytedance are two of China's most successful tech companies. Meituan Dianping started as a group buying site for food and beverage deals but has since expanded into other verticals like travel and entertainment. Bytedance, on the other hand, is the company behind TikTok. Both companies have experienced explosive growth in recent years and are now among the most valuable startups in the world. Their success is a testament to China's ability to create homegrown tech giants that can compete on a global stage. Tiktok has built one of the best and most successful content recommendation algorithms ever. Alibaba's customer service chatbot, AliMe, uses natural language processing to understand and respond to customer queries. It can handle over 90% of customer inquiries without human intervention, allowing for quick and efficient responses. Another example is China Merchants Bank, which uses facial recognition technology to identify customers and provide personalized services. AI-powered recommendation systems are also being used by companies such as JD.com to suggest products based on customers' browsing and purchasing history. Furthermore, AI-powered voice assistants, like those developed by iFlytek and Baidu, are being used to help customers with tasks such as ordering food, booking hotels, and making payments. By leveraging AI technology, Chinese companies can provide customers with faster, more personalized, and more efficient service, ultimately leading to greater customer satisfaction and loyalty.
Chinese Government Investment in AI. China has become a hotbed for entrepreneurship in recent years, with startups popping up in every industry, from e-commerce to healthcare. The government has made it a priority to encourage innovation and entrepreneurship through initiatives like the "Made in China 2025" plan. The rise of angel investors and venture capitalists has also made it easier for entrepreneurs to raise funding. However, the market is still highly competitive, and success is far from guaranteed. The government has set ambitious targets for the industry, including the goal of making China the world's primary AI innovation center by 2030. China is expected to double its investment in AI to $27B by 2026, a simply astounding figure.

Business Themes

Mobile Payments and AI. Mobile payments have exploded in China, with platforms like Alipay and WeChat Pay becoming ubiquitous. China's large population and relatively low adoption of credit cards have made it an ideal market for mobile payments. Today, mobile payments are used for everything from paying for groceries to renting bikes. The convenience of mobile payments has also fueled e-commerce growth, as consumers can easily make purchases on their phones. The future of mobile payments in China looks bright, with experts predicting continued growth and innovation.
Four Types of AI. Lee categorizes AI into four types: internet AI, business AI, perception AI, and autonomous AI. Internet AI refers to algorithms that are used to power online services like search engines and recommendation systems. Business AI is used to optimize business processes and improve efficiency. Perception AI is used to analyze and understand visual and auditory data. Autonomous AI is used to power self-driving cars and other autonomous systems. Each type of AI has its own unique challenges and opportunities, and companies are investing heavily in each area to gain a competitive advantage.
The future of the Chip Industry. In December China announced an absolutely massive stimulus to the chip industry: “China is working on a more than 1 trillion yuan ($143 billion) support package for its semiconductor industry, three sources said, in a major step towards self sufficiency in chips and to counter U.S. moves aimed at slowing its technological advances.“ One of the big geopolitical challenges of today’s age is the role of TSMC in the chip industry. “‘TSMC is just absolutely critical,” says Peter Hanbury, a semiconductor specialist at the Bain & Co. consulting firm. “They basically control the most complicated part of the semiconductor ecosystem, and they’re a near monopoly at the bleeding edge.’” With TSMC located in Taiwan, the proximity to China can be concerning for the US, especially with China repeatedly ratcheting up the tensions. China’s domestic chip industry has not been able to reach the pinnacle of chip development, which is no easy feat. TSMC believes that excellence only comes from rigorous kaizen process and is not the result of larger investment dollars. Only time will tell who will win the chip wars.
Dig Deeper

January 2022 - Seven Powers by Hamilton Helmer

This month we dove into a classic technology strategy book. The book covers seven major Powers a company can have that offer both a benefit and a barrier to competition. Helmer covers the majority of the book through the lens of different case studies including his favorite company, Netflix.

Tech Themes

Power. After years as a consultant at BCG and decades investing in the public market, Helmer distilled all successful business strategies to seven individual Powers. A Power offers a company a re-inforcing benefit while also providing a barrier to potential competition. This is the epitome of an enduring business model in Helmer's mind. Power describes a company's strength relative to a specific competitor, and Powers focus on a single business unit rather than throughout a business. This makes sense: Apple may have a scale economies Power from its iPhone install base relative to Samsung, but it may not have Power in its AppleTV originals segment relative to Netflix. The seven types of Powers are: Scale Economies, Network Economies, Counter-Positioning, Switching Costs, Branding, Cornered Resources, and Process Power.
Invention. While Powers are somewhat easy to spot (scale economies of Google's search algorithm), creating them is anything but easy. So what underlies every one of the seven Powers? Invention. Helmer pulls invention through the lens of industry Dynamics - external competitive conditions and the forward march of technology create opportunities to pursue new business models, processes, brands, and products. Companies must leverage their resources to craft Powers through trial and error, rather than an upfront conscious decision to pursue something by design. I view this almost as an extension of Clayton Christensen's Resource-Processes-Values (RPV) framework we discussed in July 2020. Companies can find a route to Power through these resources and the crafting process. For Netflix, the route was streaming, but the actual Power came from a strong push into exclusive and original content. The streaming business opened up Netflix's subscriber base, and the content decision provided the ability to amortize great content across its growing subscriber base.
Power Progressions. Powers become available at different points in business progression. This makes sense - what drives a company forward in an unpenetrated market is different from what keeps it going during steady-state - Snowflake's competitive dynamics are different than Nestle's. Helmer defines three stages to a company: Origination, Takeoff, and Stability. These stages mirror the dynamics of S-Curves, which we discussed in our July 2021 book. During the Origination stage, companies can benefit from Cornered Resources and Counter-Positioning. Helmer uses the Pixar management team as an example of Cornered Resources during the Origination phase of 3D animated movies. The company had Steve Jobs (product visionary), John Lasseter (story-teller creative), and Ed Catmull (operations and technology leader). During the early days of the industry, these were the only people that knew how to operate a digital film studio. Another Cornered Resource example might be a company finding a new oil well. Before the company starts drilling, it is the only one that can own that asset. An example of Origination Counter-Positioning might be TSMC when they first launched. At that time, it was standard industry perception that semiconductor companies had to be integrated design manufacturers (IDM) - they had to do everything in-house. TSMC was launched as solely a fabrication facility that companies could use to gain extra manufacturing capacity or try out new designs. This gave them great Counter-Positioning relative to the IDM's and they were dismissed as a non-threat. The Takeoff period offers Network Economies, Scale Economies, and Switching Cost Powers. This phase is the growth phase of businesses. Snowflake currently benefits from Switching Cost dynamics - once you use Snowflake, it's unlikely you'll want to use other data warehouse providers because that process involves data replication and additional costs. Scale economies can be seen in businesses that amortize high costs over their user base, like Amazon. Amazon invests in distribution centers at a significant scale, which improves customer experience, which helps them get more customers - the flywheel repeats, allowing Amazon to continually invest in more distribution centers, further building its scale. Network economies show in social media businesses like Bytedance/TikTok. Users make content that attracts more users; incremental users join the platform because there is so much content to "gain" by joining the platform. Like scale economies, it's almost impossible to go build a competitor because a new company would have to recruit all users from the other platform, which would cost tons of money. The Stability phase offers Branding and Process Power. Branding is hard to generate, but the advantage grows with time. Consider luxury goods providers like LVMH; the older, the more exclusive the brand, the more it's desired, and every day it gets older and becomes more desired. A business can create Process Power by refining and improving operations to such a high degree that it becomes difficult to replicate. Classic examples of Process Power are TSMC's innovative 3-5nm processes today and Toyota's Production System. Toyota has even allowed competitors to tour its factory, but no competitor has replicated its operational efficiency.

Business Themes

Sneak Attack. I've always been surprised by businesses that seemingly "come out of nowhere." In Helmer's eyes, this stems from Counter-Positioning. He tells the story of Vanguard, which was started by Jack Bogle in 1976. "You could charitably describe the reception as enthusiastic: only $11M trickled in from investors. Soon after the launch, [Noble Laureate Paul] Samuelson himself lauded the effort in his column for Newsweek, but with little result: the fund had only reached $17M by mid-1977. Vanguard's operating model depended on others for distribution, and brokers, in particular, were put off by a product that predicated on the notion that they provided no value in helping their clients choose which active funds to select." But Vanguard had something that active managers didn't: low fees and consistency. Vanguard's funds performed like the indices and cost much less than active funds. No longer were individuals underperforming the market and paying advisors to pick actively managed funds. Furthermore, Vanguard continually invested all profits back into its funds, so it looked like it wasn't making money while it grew its assets under management. It's so hard to spot these sneak attacks while they are happening. But one that might be happening right now is Cloudflare relative to AWS. Cloudflare launched its low-cost R2 service (a play on Amazon's famous S3 storage technology). Cloudflare is offering a cheaper product at a much lower cost and is leveraging its large installed base with its CDN product to get people in the door. It's unclear whether this will offer Power over AWS because it's confusing what the barrier might be other than some relating to switching costs. However, there will likely be reluctance on AWS's part to cut prices because of its scale and public company growth targets.
A New Valuation Formula. Helmer offers a very unique take on the traditional DCF valuation approach. Investors have long suggested the value of any business was equal to the present value of its future discounted cash flows. In contrast to the traditional approach of summing up a firm's cash flows and discounting it, Helmer takes a look at all of the cash flows subject to the industry in which firms compete. In this formula (shown above), M0 represents the current market size, g the discounted market growth factor, s the long-term market share of the company, and m the long-term differential margin (net profit margin over that needed to cover the cost of capital). More simply, a company is worth it's Market Scale (Mo x g) x its Power (s x m). This implies that a company is worth the portion of the industry's profits it collects over time. This formula helps consider Power progression relative to industry dynamics and company stage. In the Origination stage, an industry's profits may be small but growing very quickly. If we think that a competitor in the industry can achieve an actual Power, it will likely gain a large portion of the long-term market. Thus, watching market share dynamics unfold can tell us about the potential for a route to Power and the ability for a company to achieve a superior value to its near-term cash flows.
Collateral Damage. If companies are aware of these Powers and how other companies can achieve them, how can companies not take proactive action to avoid being on the losing end of a Power struggle? Helmer lays out what he calls Collateral Damage, or the unwillingness of a competitor to find the right path to navigating the damage caused by a competitor's Power. His point is actually very nuanced - it's not the incumbent's unwillingness to invest in the same type of solution as the competitor (although that happens). The incumbent's business gets trashed as collateral damage by the new entrant. The incumbent can respond to the challenger by investing in the new innovation. But where counter-positioning really takes hold is if the incumbent recognizes the attractiveness of the business model/innovation but is stymied from investing. Why would a business leader choose not to invest in something attractive? In the case of Vanguard competitor Fidelity, any move into passive funds could cause steep cannibalization of their revenue. So in response, a CEO might decide to just keep their existing business and "milk" all of its cash flow. In addition, how could Fidelity invest in a business that completely undermined their actively managed mutual fund business? Often CEOs will have a negative bias toward the competing business model despite the positive NPV of an investment in the new business. Just think how long it took SAP to start selling Cloud subscriptions compared to its on-premise license/maintenance model. Lastly, a CEO might not invest in the promising new business model if they are worried about job security. This is the classic example of the principal-agent problem we discussed in June. Would you invest in a new, unproven business model if you faced a declining stock price and calls for your resignation? In addition, annual CEO compensation is frequently tagged to stock price performance and growth targets. The easiest way to achieve near-term stock price appreciation and growth targets is staying with what has worked in the past (and M&A!). Its the path of least resistance! Counter-positioning and collateral damage are nuanced and difficult to spot, but the complex emotions and issues become obvious over time.

Dig Deeper

November 2021 - Ender's Game by Orson Scott Card

This month we check out the futuristic sci-fi war drama, Ender’s Game. While the book is meant for kids, its a quick read and a great story.

Tech Themes

The Metaverse. During Ender's time at Battle School, he interacts with the Mind Game, an individual game reflecting the thoughts and experiences of each person. Later, as he preps for battle, Ender uses a simulator to learn and practice commanding an army of battleships. These experiences in the simulator are completely personalized, driven by a supercomputer that can do whatever it wants to serve up experiences in the game: "You don't understand, sir. Our Battle School computer is only a part of the IF network. lf we want a picture, we have to get a requisition, but if the mind game program determines that the picture is necessary--it can just go take it." These hyper-personalized mind game experiences are similar to the latest ideas surrounding the Metaverse. The Metaverse is an unclear vision of cyberspace where individuals can interact in virtual reality, mixed reality, or augmented reality in a new computing paradigm. Facebook was so excited about the Metaverse that the company announced it was going to invest $10B in building out its virtual reality platform and changed its name to Meta Platforms, Inc. Matthew Ball has covered the Metaverse since 2018 and has penned his own definition: "The Metaverse is a massively scaled and interoperable network of real-time rendered 3D virtual worlds which can be experienced synchronously and persistently by an effectively unlimited number of users with an individual sense of presence, and with continuity of data, such as identity, history, entitlements, objects, communications, and payments." This is reminiscent of the world we explored in our September 2019 book, Ready Player One, and somewhat similar to the 1990s promise of the information superhighway. It will be interesting to see how the Metaverse develops in the coming years.
Anonymity on the Internet. As Ender continues his training at Battle School, Peter and Valentine hatch a plot to create division throughout the world. The two decide the best way to take over the world as young, intelligent children is to write blog posts under a pseudonym and gain a mass following, eventually exercising their political influence. To avoid suspicion, Peter and Valentine switch emotional positions and take on the roles of historical figures aligning with their viewpoints. Peter becomes John Locke, a liberal philosopher, and inventor of the Social Contract, while Valentine becomes Demosthenes, an Athenian hellbent on inciting a war against Macedonia. While the idea of two teenage children starting war by writing on the internet is comical now, the specter might have been possible in the pre-mass internet era of 1985, the year Ender was published. This also raises the contentious shield of anonymity offered by the internet. While some argue that complete anonymity could mean the end of rational society, others say that anonymity must be preserved. This concept of anonymity is extended further in the over-hyped decentralized, crypto/web3 world of the future, where 15-word recovery phrases might become the norm for ultimate secrecy. Internet security and anonymity are likely to evolve if we move to a decentralized computing world - whether this is good or bad remains a matter of view.
Technology and Governments. The International Force (IF) is a space army designed by the world to fight against the evil Buggers. The surprising thing about this International Force is how it unifies different governments: "Val, it was bound to happen. Right now there's a vast international fleet and army in existence, with American hegemony. When the bugger wars are over, all that power will vanish, because it's all built on fear of the buggers. And suddenly we'll look around and discover that all the old alliances are gone, dead and gone, except one, the Warsaw Pact." The Warsaw Pact was the agreement between the Soviet Union and several neighboring states following the creation of NATO. Funnily enough, the Warsaw Pact disbanded in 1991 with the fall of the Soviet Union, six years following the publication of Ender's Game. After Ender defeats the Buggers, the world immediately descends into political chaos until Peter comes to power. A once unified world with incredible technology like real-time technological communication through the Ansible is now torn apart by politics. These events bring up the broader role of government in the technological landscape. As we saw earlier this year, global non-US tech superpowers like Bytedance (owner of Tiktok) can cause immense political tension. Furthermore, companies like Taiwan Semiconductor (TSMC), that offer a unique product in a politically contentious region can even provoke the potential for war. Technology enables globalization while also raising the question of who owns non-physical products - the government, a company, or the world?

Business Themes

self-managed teams - four types of teams-1.png

Lonely at the Top. Card paints a world where the entire universe's future lies on children's shoulders. Ender becomes commander of the International Fleet, put in the challenging position as leader of older kids. He has to generate empathy while maintaining command. But Ender is just an intelligent child, and throughout the book, he finds himself in bad situations. He eventually grows to be the leader of his launch group and then the leader of his own Dragon Army. As Ender gains in stature, he loses touch with his friends. In one instance, he fears battle school enemies might jump him in the hallway and chastises Petra when she asks him to chat: "'Petra, if you had actually taken me aside just now, there are about a dozen boys following along who would have taken me in the corridor. Can you tell me you didn't notice them?' Suddenly her face flushed. "No. I didn't. How can you think I did? Don't you know who your friends are?" Many CEOs describe that the job can be lonely because you are naturally the final decision-maker. Even as a young child, Ender was forced to become a leader and suffered the mental instability of the job.
Sending a Message. The IF chooses Ender because he is a mix of his two siblings, Peter, who represents extreme violence, and Valentine, who represents empathy. Violence is a recurring theme throughout the book - personal violence between individuals, violence between nations, and violence between civilizations (humans and buggers). In two dramatic sequences, older boys try to corner an unsuspecting Ender. Ender uses his brains to evade an attack but severely injures the attacker to send a message: "They were all wondering if he was dead. Ender, however, was trying to figure out a way to forestall vengeance. To keep them from taking him in a pack tomorrow. I have to win this now, and for all time, or I'll fight it every day and it will get worse and worse. Ender knew the unspoken rules of manly warfare, even though he was only six." Ender thought he needed to message all of his potential attackers. However, these beatings weigh on him constantly, and he spends the rest of his life regretting them. The violent nature of these attacks is reprehensible and difficult to compare to the business world. But it does raise how some executives act with emotion to humiliate or denigrate employees. Recently the CEO of online mortgage startup Better.com fired 900 people over a zoom video call. Beyond the act, the message it sends to employees is even worse. These events can follow executives, with media coverage continuing for over five years after the event itself. Actions send messages. They should be taken with caution when emotion or retaliation is involved.
Self-Managed Teams. Ender is a tactical magician and completely changes the Battle Game. Ender's approach is novel: "He had the army drill in eight-man toon maneuvers and four-man half-toons, so that at a single command, his army could be assigned as many as ten separate maneuvers and carry them out at once. No army had ever fragmented itself like that before, but Ender was not planning to do anything that had been done before, either. Most armies practiced mass maneuvers, performed strategies. Ender had none. Instead, he trained his toon leaders to use their small units effectively in achieving limited goals. Unsupported, alone, on their own initiative." This approach is called Self-Managed Teams. The autonomy offered by allowing individuals to manage themselves gives extreme ownership to employees. Self-Managed teams work well in places with repeated work, where employees trust each other and have high self-awareness. This exciting concept has worked well in several businesses, including Facebook and Google.