Vol.220 Finding the Big Fish: From Meituan to J&T, from Beijing to Jakarta
Summary
- 屈田’s stock-picking method is not to chase the hottest sectors, but to lock in “timing, place and people” at the same time. He first confirms a trend that can last more than 10 years, then finds the right entry point and the strongest team, only acting when he has “70% confidence.” For early-stage investing, he may bet solely on people he knows well—but only with a small check; to make a large bet, both the “model and the people” must work. He particularly favors heavy businesses because “heavy itself is a competitive advantage,” screening out giants that only want businesses with high capital returns.
- Meituan won on cash flow and organizational strength, not its ranking in the fundraising market. During the 2011 investment decision, Meituan had fallen from No. 2 to No. 3 in group buying; Alibaba still held the valuation flat while cutting its proposed investment from $43M to $35M, completing a roughly $50M Series B with Sequoia. The key judgments were that Meituan was profitable with strong cash flow, 王兴 could compensate for his lack of offline-management experience during a crisis, and the company could develop young managers such as 沈鹏 internally. Refusing to blindly pursue an IPO after the investment and starting with an audit was another decisive move in the group-buying war.
- J&T turned globalization from a slogan into a repeatable path. When 屈田 arrived in Indonesia in 2017, J&T operated in only one country and ranked No. 2 locally; by 2019 it was already Southeast Asia’s largest, later taking the top spot in most countries with a share of roughly 35%-36%, before entering China, Latin America and the Middle East and laying out plans for the US and Europe. The program cited slightly different figures for its current market cap, but all put it in the hundreds-of-billions range; 屈田’s bigger call is that “J&T will eventually be a $100B company”—though it may take 5 years, or it may take 10.
- There is no absolute safety for China’s mid-sized platforms. Vertical companies tied to industry or hardware still have room to win, but AI Agents could bypass transaction platforms and place orders directly; even if giants cannot eliminate them, they can take away incremental growth and weaken pricing power. “Most of them are simply businesses the giants haven’t gotten around to,” so the real long-term defense is not to defend only China, but to become a global leader with deep localization in major countries.
- The 3 AI giants are playing different hands. ByteDance is betting on a closed, Doubao-style consumer product, with user scale and globalization as its advantages; even if Tencent’s foundation models lag, WeChat, IM, mini programs and open Agents could still make it the distribution gateway; Alibaba’s strongest assets are cloud, compute and To B infrastructure, allowing it to become a token provider compatible with multiple models. 屈田 expects “model intelligence to become abundant,” with roughly 90% of demand potentially met by open-source models; the decisive question may not be who owns the strongest foundation model, but who controls the gateway, compute and open ecosystem.
- Going abroad cannot be an escape pod after the domestic business loses momentum. 屈田 rejects “if you do not go abroad, you are out,” offering his own version: “if you do not have the ability to go abroad, you will be out faster.” Companies should “fix the roof while the sun is shining”: first build China-champion-level products, organizations and technology, then choose a region where they can win decisively; overseas operations must also generate their own cash and cannot rely on headquarters funding indefinitely.
- Regional selection comes down to whether you can win the first battle. The US and Europe are better suited to technological innovation, novel products and consumption upgrades, but carry high geopolitical uncertainty; Southeast Asia is better for mass-market consumption, and 屈田 sees especially strong long-term policy certainty for Chinese companies. Temu proved that a weaker platform can prise open the US market, but pure cross-border commerce will hit local-infrastructure bottlenecks; TikTok commerce is more important because it first built deep localization in Southeast Asia, where its market share has reached 70%-80% of Shopee’s, and 屈田 believes it could overtake Shopee this year or next.
- ATM is choosing long-duration capital, ecosystems and ground troops—not hot deals with which to prove itself. 屈田 acknowledges that focusing overseas means missing some opportunities in China EVs, AI and embodied intelligence, but describes the trade-off as “give up big to gain big, bet big to win big.” He does not want to follow star projects such as OpenAI in the US; he wants to be the “first investor” or overseas co-founder of global companies. Even after J&T’s listing, he says he “has not sold a single share.” The end goal is to build an evergreen fund backed by multiple global champions.
Deep dive
1. “Just for Fun” Shaped 屈田’s Way of Choosing
屈田 traces his starting point to his family. His father entered Xi’an Jiaotong University from a rural village in Shaanxi and later helped draft “Made in China 2025”; with several generations of strong mathematicians in the family, 屈田 developed an early instinct for logic and technology.
He encountered the Apple II and learned programming in third or fourth grade, but found coding dull. He preferred physics, which connected abstract mathematics to the real world. “If I forget a formula, I can derive the one I need on the spot”—that habit of deriving conclusions from first principles later carried into his investing.
More than 80 students entered Peking University’s physics department in the class of 1996; only 4 or 5 were still doing physics years later. By his sophomore year, 屈田 knew he would not spend his life researching physics. Looking back on his choice of major, he borrowed a line from the Linux founder’s autobiography: “Just for fun.”
2. Peking University Was a Training Ground for Hunting Opportunities
Around 1997, the internet began to emerge. 屈田 learned HTML and built websites for 5-star hotels listed in China Yellow Pages. He scanned the Huanglong Hotel’s brochure, entered the copy and earned RMB1,000—roughly 6 months of living expenses at the time.
He set up a ticket-resale booth on campus, founded a film society and invited 贾樟柯 and 王小帅 to Peking University. Most of these collaborations began with cold calls. “I said I was from Peking University and wanted to work with you”—the point was to act on his interests rather than wait for permission.
Football worked the same way. After losing repeatedly in the knockout rounds of the Peking University Cup, he decided the tournament format was unfair and organized his own science-student football league. The point was not the campus anecdote, 屈田 says, but seeing a gap and changing the rules.
3. Sina, Tencent and Yahoo Showed Him Content, Products and Global Companies
During the 2000 internet winter, Sina had no public openings. 屈田 contacted its news center directly and offered to work part-time without pay; only after the channel head approved did the company create a slot. What he liked most at Sina was preparing background material in advance, then publishing exclusive reports as soon as a story broke.
When Tencent’s website division recruited him in 2003, he worked in product operations rather than as a pure product manager. For the first time, he saw the power of a product manager to coordinate technology, resources and a virtual team. “Product managers are the cradle of CEOs”—the idea made an ambitious 屈田 want to move fully into products.
After less than a year at Tencent, he joined Yahoo China at the end of 2004. The logic was straightforward: Sina and Tencent were already China’s leading internet companies, so he wanted to work for the largest US internet company of the time. 周鸿祎 was then president of Yahoo China.
4. Yahoo’s Merger with Alibaba Revealed the Power of Culture and Capital
When Yahoo China merged with Alibaba in 2005, 屈田’s first reaction was: how had Yahoo been acquired by this orange, unfamiliar Alibaba? But 马云’s speeches, the chartered trains to Hangzhou and the integration of the teams quickly convinced him that Alibaba was a strong company the market had underestimated.
Alipay had only 40 or 50 people at the time, with 陆兆禧 personally leading the interface with Yahoo. The “deep feeling and brotherhood” at gatherings by the lake reappeared more than a decade later in J&T’s founding team. 屈田 saw that emotional bond as organizational cohesion, not just a drinking-table story.
He studied the transaction structure in depth. Yahoo received roughly 40% of Alibaba, SoftBank about 30%, the Alibaba team around 20%, with other investors holding the balance; SoftBank was also a key force behind the deal. The acquisition was 屈田’s first systematic study of 孙正义 and shifted his interest from products to the shareholder perspective.
屈田 agrees with 孙正义’s principle of acting with 70% confidence, but rejects reducing him to a gambler. The team had already studied China’s leading B2B projects; 孙正义’s quick decision after meeting 马云 looked more like doing the homework, then stepping in to negotiate the deal and secure the stake.
5. Being a Shareholder Was Closer to the Freedom He Wanted
After actually working as a product manager, 屈田 found that he preferred constantly encountering new things. An investor could be part co-founder while retaining some freedom. The difference between being a professional manager and being a shareholder became the fundamental reason he turned toward investing after 2005.
China’s venture-capital industry was still in its early stages, and mainstream hiring favored Goldman Sachs, McKinsey and US MBA backgrounds. With a physics degree, 屈田 enrolled in Tsinghua’s MBA program while working, focusing on corporate finance and modeling.
Tsinghua’s venture-capital course was taught by KPCB venture partner Peter Liu and brought in practitioners such as 周志雄 and 邓锋. An then-unknown young entrepreneur also attended: 王兴, who had just sold Xiaonei and was looking for his next project.
That classroom meeting and exchange of contact details laid the groundwork for the Meituan investment 5 years later. 屈田 did not suddenly “succeed in a career switch”; he built an investable range layer by layer through interests, formal training, industry relationships and internal projects.
6. The KuXun Due Diligence Was His Real-World Exam for Alibaba’s Strategic Investment Team
When Alibaba researched an investment in KuXun, 屈田 joined the business and technology diligence as a product manager for web search. He had previously conducted a background check on Mop.com for BVP and knew how to structure the report, producing material that left a strong impression on 张飞燕.
KuXun was founded by 陈华 and 吴世春; 张一鸣 had served as its head of technology, which also led to a relationship with SIG. 屈田 did not present the episode as a successful investment, but as his first entry into a node of the founder network.
When the corporate-development team expanded from 张飞燕 alone in 2007, it needed someone who understood internet products and technology and knew Alibaba from the inside. 屈田, who had just completed his MBA, fit the profile and formally moved from product management into corporate development and strategic investment.
7. Alibaba’s Early Strategic Investments Served Business Needs and Long-Term Strategy
Around 2008, corporate development was upgraded into the strategic-investment division, with roughly 10-plus people given a $300M budget. 张蔚 oversaw the group, while 张飞燕 and 谢世煌 served as partners; 屈田 was the person doing most of the project-level work and taking the lead on individual deals.
The first pipeline was business-led, seeking assets around traffic, revenue, data and ecosystem value for Taobao, Alimama, Alipay and Alibaba Cloud. Business units were nominally the sponsors, but the strategic-investment team often sourced targets from the market and then sold the opportunity internally to the business head.
The second pipeline was driven by strategy and financial returns, centered on mobile internet and cloud computing and decided by the investment committee, which included 马云, 曾鸣 and the investment partners. Meituan, Sogou and UCWeb were examples of initiatives not originated by business units.
8. Small Acquisitions Left Assets—and Lessons Beyond Price
Alibaba acquired PHPWind in 2008 and later recommended its team to Alibaba Cloud. Its founder, 王学集, went on to found Tuya Smart. The value of such projects was not only the forum software, but also the talent flywheel created by connecting teams with future founders.
In 2010, Alibaba Cloud acquired CNZZ, a website-analytics tool, from 蔡文胜 for roughly $20M. The decision rested with 王坚, and 屈田 considers it a classic combination of business, data and cloud infrastructure.
The sharper counterexample was 91 Mobile Assistant. Alibaba could have acquired it at a valuation of roughly $20M, but walked away after much of the technical team departed; Baidu later bought it for $1.9B. 屈田 later joked to the company’s head: “Luckily I didn’t acquire you back then—otherwise there would have been no ‘afterward.’”
9. Alibaba Mapped the Mobile Internet Before the Smartphone Boom
屈田 had worked on Yahoo’s wireless products and participated in 3G tests with Nokia, China Mobile and Huawei. Just after moving into investing in 2007, he argued that “the mobile internet would arrive sooner or later” and began looking for the critical entry points that would differ from the PC era.
In 2008 and 2009, the investment team scanned roughly 200-300 mobile-internet companies, mapping the stack from hardware, operating systems and middleware to applications and the cloud. The exercise sought opportunities while checking whether a mobile-commerce player could emerge to disrupt Taobao.
Its killer-app calls were not all right. The team thought Baidu would probably retain text feeds, Tencent would retain IM and Alibaba would retain e-commerce, but did not anticipate Toutiao, Pinduoduo or Douyin commerce. It saw opportunities in images and video, but not the specific forms that would become Xiaohongshu and short video.
The more accurate calls were local services, maps and transportation. Before group buying had emerged, the team had already placed Dianping-style services, maps and ride-hailing on its roadmap. Once companies surfaced, Alibaba could follow the map rather than learn from scratch after a trend became obvious.
10. Alibaba’s 2009 Wireless Strategy Meeting Turned the Thesis into 3 Actions
In March 2009, Alibaba Group required Taobao, B2B, Alipay, Alibaba Software and other core businesses to present their wireless strategies. The strategic-investment team also submitted a group-wide view, with 屈田 drafting most of the presentation; 于永福 of UCWeb was invited to share his perspective.
马云 ultimately reduced the complicated discussion to 3 actions: first, go all-in on Alipay’s “card pass,” later known as Quick Payment; second, go all-in on Mobile Taobao and “build another Taobao on the phone”; third, make a strategic investment in UCWeb.
UCWeb was valued at roughly $70M at the time, and Alibaba acquired about 10% in the first round, continuing to add until it bought the company. 屈田 liked not only its traffic but also the quality of its cash flow as an early mobile company. Having lived through the 2000 winter, he was never optimistic about companies that required continuous fundraising.
When Mobile Taobao first began gaining traction, perhaps 20% of its traffic came from UCWeb. But 屈田 saw the longer-term value in the people: UCWeb alumni later contributed to Gaode, Alibaba’s entertainment businesses, Alibaba Health, gaming and Qwen. “Compared with the product itself, acquiring the talent was more important.”
11. Group Buying Created the First Online-to-Offline Transaction Loop for Local Services
After Groupon emerged, Alibaba saw it as more than advertising for local services. It moved the transaction online while fulfillment happened offline, fitting Alibaba’s preference for a closed transaction loop. 屈田 therefore spoke with every major group-buying company.
When 屈田 asked 王兴 why he was not continuing to build a social network, 王兴 answered that Xiaonei, Hainei and Fanfou connected people to one another, while Meituan connected people with local merchants and services. Meituan thus became a broader connection thesis rather than a Groupon clone.
Lashou was No. 1, Meituan was about 70% of Lashou’s size and Manzuo was No. 3. After 55tuan used acquisitions to move rapidly into second place, Meituan slipped from No. 2 back to No. 3 before Alibaba’s final decision. The project ran from October 2010 to June 2011, roughly 8 months.
12. Alibaba Rejected Lashou Because of Cash Flow and Organization, Not Valuation
Lashou had just raised money at a valuation of roughly $220M and said it would raise $100M-$200M at a $1B valuation, while offering Alibaba the chance to invest $20M at the old valuation. For the strategic-investment team, it looked like an attractive mark-to-market opportunity.
But Lashou demanded a non-compete and later refused to allow Alibaba to conduct full diligence. 屈田 denies that 王兴 ever proposed a similar condition and considered “not allowing diligence” a direct red flag: insufficient information means no large bet.
The operating data also favored Meituan. Lashou was larger but deeply loss-making and burning cash, while Meituan was profitable with strong cash flow. 屈田’s underlying preference was explicit: “I don’t think fundraising markets will always remain open.” Cash-flow survival came before growth.
In the final meeting, Lashou’s founder proudly described a “wolf-pack management” system: the 10 regional divisions had no assigned territories and competed against one another for business. Alibaba concluded that a team assembled around IPO promises and financial incentives could only fight in a favorable environment, like 苻坚’s army at the Battle of Feishui—“once it encountered a setback, it would scatter quickly.”
13. 王兴’s Adaptability in Crisis Changed 屈田’s Assessment
Before investing, 屈田’s view of 王兴 was simply that he was smart but difficult to communicate with. During the fundraising process, he watched 王兴 improve at extraordinary speed and formed a generational judgment: among founders in their early 30s, 王兴 was the strongest he had seen, and “10 years is roughly one generation gap.”
55tuan treated Meituan as its main rival and poached 13 or 14 of its regional heads. 王兴 had originally disliked dining and drinking with offline managers, but in the crisis he spoke with them one by one and stabilized the team. 屈田 saw that he could take on work he did not naturally enjoy.
屈田 used the phrase “a gentleman is not a tool” to describe that capacity. 王兴 was not limited to product and technology; he had significant potential and room to adapt. “When it is truly necessary, he can step up”—more important than any particular skill he already possessed at the time.
王慧文, 郭爱怀, 慕容君 and 陈亮 were young, capable and tightly bonded, mostly classmates and long-term partners. Compared with the founders of Lashou and 55tuan, Meituan also understood mobile internet earlier and was among the first to launch a mobile app, creating a double fit between team and direction.
14. 沈鹏 Gave Investors Evidence of Meituan’s Ability to Grow Middle Management
屈田 brought in 甘家伟, a veteran of Alibaba’s B2B business, for joint diligence. His main concern was that Meituan was still weak at managing offline teams. That happened to be an Alibaba strength, so the plan before investing was to export an offline-management system to Meituan.
沈鹏, who had only recently graduated, was already running the Tianjin region and may have had hundreds of people reporting to him. To 屈田, who had observed organizations at Sina, Tencent, Yahoo and Alibaba, this showed that Meituan could identify and promote young managers from within.
His judgment was that externally hired professional managers tend to be less stable; a durable company must continually grow new talent. Even if several regional heads were poached, a company that could produce another 沈鹏 would not be broken by one round of raids. That was harder evidence of organizational strength than founder charisma.
15. Meituan’s Real Series B Decisive Moves Came After the Investment
Alibaba had initially planned to invest $43M. After Meituan’s ranking deteriorated, Alibaba kept the valuation unchanged but cut the investment to $35M, completing a roughly $50M Series B with Sequoia. 屈田 saw this as reducing exposure rather than overturning the long-term thesis.
The first board meeting discussed whether Meituan should follow Lashou and 55tuan toward an IPO. Alibaba’s strategic-investment team judged that both businesses lacked substance and were likely to fail in the public markets. Meituan therefore would not file immediately, but would first hire one of the Big Four auditors and remain in contact with investment banks so it could move quickly if a rival actually listed.
Before the investment decision, 马云, 蔡崇信 and other investment-committee members received an anonymous email accusing Meituan of fabricating data; the 3 people directly running the project did not. 屈田’s counter-observation was that rivals’ detailed knowledge of Alibaba’s internal process—and their fear of Meituan—was itself a signal.
Lashou and 55tuan poached employees with high salaries, options and IPO promises. When the IPOs failed, morale and organization collapsed rapidly. Meituan avoided the damage of a failed listing and ultimately emerged from the group-buying war. 屈田 calls the early audit, delaying the IPO and bringing in 甘家伟 the decisive moves in the eventual outcome.
16. Leaving Alibaba Was About Control Over Decisions and Economics
屈田 liked strategic and financial investing but became increasingly convinced that “a person gets only a few truly good investment opportunities in a lifetime.” If he could not make the call, an organization might miss the opportunity; even if it captured one, the question was how much of the upside a professional manager could share.
In 2012, he joined CDH Investments because, although he knew how to run deals, he had not experienced the full cycle of a market-based fund—fundraising, investing, managing and exiting. CDH, whose record included Mengniu and Nanfu, was then one of China’s top private-equity firms and offered the institutional operating experience he lacked.
After leaving in 2014, he founded Bat Capital with former colleagues from Alibaba and Tencent and other friends. The name came from BAT. The initial plan was to back employees leaving large companies to start businesses, but the mobile-internet cycle was already late and the opportunity set smaller than expected. The strongest investments, including Tuya Smart and Waterdrop, still came from the continuation of long-term relationships.
17. Alibaba Did Not Passively Receive a Ticket to the Mobile Internet; It Built One Early
李翔’s initial impression was that Tencent transitioned smoothly through WeChat while Alibaba’s path was more precarious. 屈田 offered the opposite view: Alibaba was a strategy-driven company that had systematically prepared for mobile internet since 2009, while WeChat also benefited from the serendipity created by 张小龙’s product genius.
Alipay Quick Payment, Mobile Taobao, UCWeb, Gaode and the cloud together formed Alibaba’s mobile infrastructure. 屈田 emphasized that WeChat Pay’s later ability to connect rapidly with banks also benefited from Alipay having already run the same bank-partnership playbook.
Baidu tried to fill its mobile-entrance gap by buying 91 Mobile Assistant for $1.9B, with a result different from Alibaba’s and Tencent’s. 屈田 used the episode to praise Google’s foresight: it acquired Android early and built it into a dominant open-source operating system, while using YouTube to secure video.
Alibaba’s Aliyun OS strategy was extinguished by Google’s Android strategy before it could mature, showing that the entry layer is not an ordinary application market. 屈田’s postmortem is that mapping the industry chain early does not guarantee every step will work, but it lets payments, traffic, maps, cloud and talent form a portfolio through a cycle change rather than betting on one product.
18. The Long-Running Meituan-Alibaba Conflict Began When Logistics Re-entered Local Services
李翔 cited Meituan’s investments in Guming, Mixue Bingcheng, Unitree and Li Auto. 屈田 said life services were areas Meituan should invest in because it owned traffic and data, while its technology investments reflected 王兴’s sensitivity to the frontier of technology and to technical founders.
Alibaba initially understood Meituan as “e-commerce without logistics”: the transaction happened online, while the service was consumed offline. Food delivery introduced instant delivery, turning the business back into logistics and creating competition with e-commerce infrastructure.
Alibaba once explored whether it could acquire Meituan; 王兴 signaled his desire for independence by saying “we are still young.” After Meituan entered food delivery, Alibaba tried again and was rejected, then turned to Ele.me. 屈田 therefore sees today’s food-delivery and e-commerce rivalry as predictable—a conflict over business boundaries that was planted years ago.
19. AI Transformation Starts by Answering: What Is the Core Gateway?
The lesson 屈田 takes from the previous transition is not to copy mobile-internet tactics, but to first ask whether the company’s core is To C or To B. To C competes for the gateway and traffic; To B competes through cloud, compute and enterprise infrastructure.
Alibaba has long tried to fill its consumer-side gap, and investments in mobile operating systems and Youku can be understood in that context. But its more durable asset is its To B cloud business. In AI, it can still build a foundational-services advantage through cloud, chips, data centers and tokens.
The device itself has not been settled. Phones may continue to evolve, while earphones and glasses could also carry voice interaction. 屈田 is unsure whether the phone will remain the only gateway, but certain that “staring at a screen all day is not very healthy.” Products positioned as enemies of the phone have historically failed, though AI may change the interface.
20. Closed Doubao and Open “Lobsters” Represent 2 AI Distribution Structures
屈田 divides consumer AI into 2 categories: closed apps in the ChatGPT or Doubao mold, and open “lobster” architectures that can connect to different models and continuously generate 2.0 and 3.0 versions. The key question is whether most users ultimately remain inside one closed product or move toward open Agents.
ByteDance’s strength is the huge user base and token consumption brought by Douyin. Doubao could win in China and globally through distribution; its weakness is that chatbot inference is expensive and the monetization loop is difficult, especially with low willingness to pay among Chinese users. That makes it hard for the platform to keep allocating its most expensive compute to a free app.
Tencent’s opportunity may not be to copy Doubao, but to wait until open Agents stabilize and then gradually open WeChat, IM, mini programs, social graphs and information under security constraints. “Connecting WeChat to Agents” could make Tencent a major gateway for lobster-style products.
ByteDance’s other card is globalization: its consumer capabilities overseas are clearly stronger than Tencent’s. 屈田 even believes over the long term that ByteDance could become the world’s largest e-commerce company, because China accounts for only about 20% of the global economy and competition overseas is less intense than in China.
21. Alibaba’s AI Edge Is Compatibility; Tencent’s Is the Gateway, Not Model Ranking
屈田 compares 吴永明’s “A A T H” strategy to the creation of Alimama. The bottom layer would provide tokens, compute and data centers like Alibaba Cloud, while the top layer could work with different models and service providers, especially multiple open-source models.
The “lobster” boom has left Chinese giants broadly short of compute, highlighting Alibaba’s infrastructure base. Even if customers build their own foundation models, they still need power, data centers and chips. Alibaba can therefore be a service provider without betting on one model.
李翔 asked why Tencent’s Agent would not be built on its own model if its foundation model was already behind. 屈田’s answer was that if the gateway can call external models such as Qwen, the foundation model may not determine the winner. In an extreme case, “Tencent plus Alibaba could compete with ByteDance.”
22. AI Agents Could Bypass Transaction Platforms and Reallocate the Profit Pool
屈田 still sees opportunities in industry- or hardware-linked companies such as Meituan, Didi, Ctrip and Xiaomi. But if an Agent can understand demand and execute a transaction directly, the existing search, comparison and checkout gateways of platforms could be bypassed by an upper-layer agent.
The Doubao phone assistant placing orders directly on behalf of users has already demonstrated the risk. 李翔 summarized it as “skipping the transaction platform,” and 屈田 agreed. What happens to transaction platforms once Tencent’s Agent is connected to services remains unknown.
The more realistic threat may not be that companies disappear, but that they lose incremental growth and pricing power. “Your 80% market share and 30% market share may both come with rising GMV,” but margins will be completely different. In China’s intensely competitive market, no one is absolutely safe.
23. Going to Indonesia in 2017 Meant Leaving a Mature Mobile-Internet Market
屈田 describes his first 2 major career choices as entering the internet in 2000, then shifting into investing in 2007 and backing mobile internet and cloud. By 2017, early financings for major opportunities such as Douyin, Pinduoduo and Xiaohongshu had largely been completed; staying in familiar sectors offered little chance of capturing the same kind of upside.
Digital currencies, hard tech and AI were hot, but their technology timelines were unclear and top funds had been investing in them for years. He was looking for a new space that was “large enough, long enough and early enough,” not a chance to catch up in an already-established sector.
Friends at international funds invested in GoJek, Alibaba acquired Lazada, and Tencent, JD.com, Meituan and Ant were all moving into Southeast Asia. More important, China had moved from copying US models to exporting original models such as ByteDance, Xiaohongshu and Pinduoduo.
His long-term logic was that global companies require talent, technology, organization and 10 or 20 years of accumulation. China had previously lacked enough of those assets, but by 2017 the conditions were beginning to exist. “After investing in China champions, the more challenging task is investing in world champions.”
24. J&T Turned “Born Global” from a Vague Idea into a Repeatable Sector
When he first arrived in Southeast Asia, 屈田 wanted to take China’s leading companies abroad, but found that they still believed China offered plenty of room to grow. He then looked at local founders, but saw a gap versus the teams at Meituan and UCWeb, while most could operate in only one country.
J&T changed his view. An entrepreneurial ecosystem formed by Chinese founders already existed overseas. When 屈田 met the team in 2017, J&T operated only in Indonesia and ranked No. 2 locally. The upper bound he could imagine at the time was becoming Southeast Asia’s largest express-delivery company.
By 2019, J&T’s overall scale was already the largest in Southeast Asia. It later reached No. 1 in most countries with a regional share of roughly 35%-36%, then entered China, Latin America and the Middle East and began laying out the US and Europe.
The path showed that an Indonesian company did not have to remain in Indonesia, and a Southeast Asian company did not have to remain in Southeast Asia. 屈田 consequently opened up the “born global” investment category and gained credibility and leverage in overseas discussions with later Chinese giants.
25. A World Champion Must Truly Be a Leader in Major Countries
屈田’s definition of a “world champion” is demanding. It is not a company that uses China’s population to stack up global revenue or users, but one that operates globally and commands high share or even a monopoly position in most major countries.
That requires global and local capabilities to work at the same time—what he calls “glocal.” He cites TikTok, OPPO and J&T: even an online product such as TikTok deepened its operations through acquisitions and local teams, while OPPO and J&T built deeper offline networks and local organizations.
Another path is moving from China champion to world champion. DJI’s global technology lead reduced direct competition, but increasing share still required offline distribution. In several countries, partners of 屈田’s team distribute DJI, helping it move from simple product sales to deep local-channel penetration.
26. Chinese Champions Going Abroad Can Recreate Private-Market Opportunities
Chinese champions are often already public companies and traditionally outside the range of VC. ATM’s approach is to separate their overseas businesses, bring in new investors and local partners, and turn the business into an investable private-market vehicle again.
ATM offers more than generic introductions. It provides 3 layers of capability: determine regional and country priorities, select the mode of entry, then put the team, channels and local relationships in place to shorten the timeline and improve the odds of success.
屈田 rejects shallow exposure to every market. The aggregate may look large, but the company would have no core advantage or brand in any country. The better path is to go deep in one region and one country, reach scale or even No. 1, then expand.
Equity investing allows ATM to share in the upside. Its ideal role is not financial adviser, but a long-term co-founder of the overseas business.
27. The First Principle of Market Selection Is Not Size, but Whether You Can Win
Whether the first target is the US and Europe or an emerging market, 屈田 starts by asking whether the company can build revenue and profit and generate its own cash. “Relying on headquarters to fund an overseas market for the long term is not credible,” because overseas operating costs are far higher than in China.
The US and Europe are better suited to technological innovation, novel products and consumption upgrades; Southeast Asia is better for mass-market consumption, where overly premium products may not sell. Product attributes, price points and existing capabilities determine where the first win is most achievable.
Companies must also choose the objective: earn meaningful revenue and profit in a large market while ranking fifth or sixth, or enter early, take the No. 1 position and enjoy the share economics over time. Strategically, 屈田 prefers the latter—decisive and durable victory.
28. 屈田 Did Not Choose a Silicon Valley China-Focused Fund Because It Was Not His Best Field
屈田 calls the path of raising money in Silicon Valley and investing in Chinese founders “a relatively crude strategy.” The strongest deals are usually taken first by mainstream Silicon Valley funds, it is difficult to predict whether Chinese founders will consistently build leading application or foundation-model companies, and he does not personally understand the local Chinese-founder ecosystem well enough.
He is not satisfied with making money by following OpenAI-, Anthropic- or SpaceX-style projects. He wants to be the company’s first investor, or at least its first overseas investor, holding enough equity to play a co-founder role. Given the uncertainty around US-China decoupling and nationality choices, he sees emerging markets as the field where he has a more certain advantage.
29. Missing China’s Hot Sectors Was the Deliberate Cost of “Giving Up Big to Gain Big”
李翔 asked whether shifting the focus overseas meant regrettably missing China’s new-energy vehicles, AI and embodied intelligence. 屈田 quoted a strategic principle he learned from 曾鸣: “Give up big to gain big, bet big to win big.”
Staying in China would mean dividing the same pool of opportunities with excellent funds, while the eventual size of that pool was uncertain. If a global company succeeds, its scale could be 3 to 5x that of a China-only company, and ATM has a harder-to-replicate advantage in that market.
He does not consider himself cut off from Chinese technology opportunities. Hard-tech companies ultimately need to go abroad, and ATM can share in their growth through overseas partnerships. Since the second half of 2024, the team has actively studied AI, robotics, embodied intelligence and energy, but will wait until the technology is closer to maturity before applying its industrial-landing capabilities.
屈田 admits he has never been particularly good at experimental frontier technology. When he invested in the internet and mobile internet, the underlying technologies were already relatively mature. His real strength is figuring out how to combine mature technology with local markets and industries, so understanding a little later and executing more deeply fits his circle of competence.
30. ATM’s Moat Is Not an Information Edge, but Overseas “Momentum”
A dinner with the head of a trillion-yuan market-cap company may allow 屈田 to understand an unfamiliar industry quickly. The harder question is why those leading companies would want to work with him. The first answer is that ATM can judge where they are most likely to achieve a durable victory.
Using the framework of the “Form and Momentum” chapter of The Art of War, he says leading companies accumulate “form” in China: products, technology, capital and organizational strength. Overseas, they still lack “momentum”—local networks, channels, teams and reliable allies.
ATM wants to be a local force whose values and interests remain aligned with the company over the long term. It can help decide whether to enter the US and Europe, Southeast Asia or another emerging market, and solve how to build the team, choose partners and establish channels.
J&T gave ATM an important lever. 屈田 says the team may disclose several more “headline cases” over the next 1 or 2 years. One success can be dismissed as luck; repeated successes create compounding ecosystem value and credibility.
31. Beyond J&T, the Most Valuable Track Record Includes Avoiding Traps and Admitting Misses
On J&T’s long-term outlook, 屈田 offered a clear call with a time caveat: “It will eventually be a $100B company, but I don’t know whether that takes 5 years or 10.” The program described its current value as nearly HK$100B at the opening and later summarized it as RMB100B; the figures were not fully consistent.
He looked at eFishery when it was valued at several million dollars and initially felt the model was not credible. Even after the company raised at a valuation above $1B, delivered rapid revenue growth and did not burn cash, he still judged it “too good to be true.” It was later shown to involve massive financial fraud.
The clear miss was Shopee. When 屈田 arrived in Southeast Asia, it had just listed with a market cap of roughly $5B. He liked the e-commerce business but did not buy the stock because he was focused on private markets. The program estimated that its market cap later reached roughly $60B-$80B.
Shopee’s founding team had deep ties to China, with substantial R&D and talent recruited from Chinese internet companies. 屈田 calls it the Southeast Asian company most like Pinduoduo. Aside from the Shopee public-market opportunity, he believes J&T remained the largest private-market opportunity in Southeast Asia after 2017.
32. The Global-Investing Framework Has Evolved from 1.0 to 3.0
Version 1.0 established only 2 points: Southeast Asia had a large digitization opportunity, and Chinese business models could be exported. 屈田 did not yet know whether the teams would come from Chinese giants, local founders or elsewhere.
Version 2.0 was triggered by J&T. It abandoned the fixation on rebuilding a large internet platform and shifted toward infrastructure, platform applications and online-offline consumer businesses, while focusing on Chinese teams building global companies from day one.
Version 3.0 is no longer confined to Indonesia or Southeast Asia, but targets global markets including Latin America, the Middle East, the US and Europe. The sectors now include technology and energy, and the team universe has expanded from overseas incubation to partnerships with Chinese champions going abroad. The one constant is the search for “future global champions.”
33. Most of the 2022-23 Anxiety About Going Abroad Came at the Wrong Time
“If you do not go abroad, you are out” was widely repeated in 2022 and 2023. 屈田 thought many companies going abroad at the time were China’s middle- or bottom-tier players, or businesses already under pressure, using overseas markets as an escape route from domestic competition.
His counter-slogan was: “If you do not have the ability to go abroad, you will be out faster.” More rational companies would make a trial run, discover they lacked the capabilities and stop. Those that forced the expansion often performed poorly because overseas costs and organizational complexity are higher.
He uses “fix the roof while the sun is shining” to describe the timing. Companies should go abroad when their domestic business is healthy and their product, technology and organization are strongest—not after the core business has already lost momentum and they need a second curve.
The true leaders only began acting seriously in 2024 and 2025. They had accumulated more complete capabilities and observed the results of OPPO, J&T, Mixue Bingcheng and Pop Mart. 屈田’s view is that the main phase of giant Chinese companies going abroad is only beginning now.
34. Pop Mart and Mixue Bingcheng Forced 屈田 to Revise His Consumer View
Looking back at Southeast Asia, 屈田 says the 2 important opportunities he missed beyond Shopee were Pop Mart and Mixue Bingcheng. Neither performed well immediately after going abroad, but if ATM had proactively partnered with them then, it might have found a way to participate.
The misjudgment came from 2 factors: insufficient understanding of the teams and underestimating whether new categories such as designer toys and milk tea could become cross-cultural hits. Unlike OPPO and J&T, which expanded smoothly at the beginning, these businesses took 2 to 3 years of adjustment to find their footing.
The revision pushed ATM to pay more attention to the “China champion to world champion” path. A weak initial showing does not mean a product lacks global potential; the patience and iteration capacity of a leading team are themselves long-term variables that must be priced in.
35. Southeast Asia’s Advantage Is Not Just Lower Competition, but Policy Certainty
屈田 believes the US, Japan and India all require companies to study the market and geopolitics at the same time. India is especially uncertain, while the US and Japan also carry substantial uncertainty.
The Middle East was relatively stable, but new uncertainty had emerged when the conversation took place. Latin America is relatively attractive but could be affected by the US’s Monroe Doctrine orientation. Africa offers more policy room, but it remains difficult to judge when its consumer market will fully mature.
Transsion and LeComfort, a consumer business under Sunmaxx, both spent more than 10 years building in Africa before seeing a breakout. Emerging markets do not deliver their upside immediately after entry; they require companies to endure a longer cultivation period.
By comparison, 屈田 sees Southeast Asia—especially Indonesia, Malaysia and Thailand—as the region with the strongest certainty for Chinese companies going abroad, given their long-standing ties with China. Competition will intensify, but remains well below China’s level.
36. Temu Proved the US Can Be Disrupted; TikTok Is Closer to a Global Platform
Pinduoduo attacked the US directly through Temu, producing results that surprised most observers. 屈田 believes 黄峥, who studied, worked and observed US consumers over an extended period, is among the Chinese internet founders with the strongest understanding of the US market; SHEIN’s earlier success also provided a blueprint.
The market structure was not what many outsiders assumed. Taobao once held 70%-80% of China’s e-commerce market, while Amazon had roughly 30% in the US. If Pinduoduo could challenge Taobao from a weak position, it might use the same method to challenge Amazon, which does not monopolize the US.
But Temu’s cross-border model is lightweight. Further growth requires overseas warehouses, fulfillment, channels and other local infrastructure. 屈田 sees TikTok commerce as having the stronger long-term position because it is deeply localized and did not start with the US, but expanded first in Southeast Asia and the UK.
The program’s estimate was that TikTok commerce had reached roughly 70%-80% of Shopee’s scale in Southeast Asia and could overtake Shopee that year or the next. Even if policy constraints limit it in the US, it could still become “the world’s largest e-commerce company” through markets outside the US.
37. Meituan and J&T Both Chose Heavy Entry Points Inside Large Trends
屈田 uses “timing, place and people” to review the 2 representative investments. Meituan’s timing was mobile internet and its entry point was in-store services and food delivery; J&T’s timing was the next phase of mobile internet globalization and the globalization of Chinese e-commerce platforms, while its entry point was express-delivery infrastructure.
Entry timing also mattered. Meituan started in 2010 and reached scale in 2011-12, just as mobile internet was accelerating. J&T entered Indonesia in 2015, as e-commerce platforms including Shopee were beginning to gain traction.
“People” refers not only to how strong the founders were at the time, but also to their capacity to learn. Meituan captured food delivery, merged with Dianping and overtook Ele.me; J&T expanded from Indonesia into Southeast Asia, then China, Latin America and the Middle East. Each step required absorbing new talent.
Both companies chose difficult, heavy businesses, which made the outcome more certain. “Why would an easy business fall into your lap?” Meituan rapidly expanded to roughly 200 cities, raising the difficulty of offline management while attacking Ele.me’s weaknesses. Heavy operations were themselves a barrier that screened out competitors.
38. The Model May Not Be the End-State Moat; Open Agents Could Reduce Its Importance
李翔’s challenge was that a large company building an Agent should logically build it on its own model; if Tencent’s foundation model was already behind, what was the logic for building an Agent? 屈田 answered with the openness of the “lobster”: it can call any model and could therefore reduce the importance of a single foundation model.
Following the Musk-style logic 屈田 cited, the long-term bottleneck behind models is compute; whoever can stack more compute may ultimately become stronger. Talent also moves between companies—“there are no secrets in Silicon Valley”—making it difficult for one model to rule the market permanently.
He went further: “model intelligence will become abundant.” Perhaps 90% of tasks can be handled by open-source models, with only roughly 10% requiring the strongest capabilities. Tencent’s real requirement is to ensure that no single model establishes an absolute monopoly, not necessarily to build the No. 1 model itself.
Tencent could also acquire a leading model company or recruit a key team. 屈田 believes the core leaders who determine a model’s quality may number only 3 to 5. Their role is not to complete all the R&D themselves, but to attract the 10 or 20 best people around them.
39. Missing Didi Exposed the Limits of 屈田’s Certainty Framework
王刚 showed 屈田 Didi at an early stage, when its valuation may have been only several million yuan. But 屈田 was at CDH, the project was too early and the check size too small, putting it outside the institution’s range from the start.
He also disliked the model of starting with taxis. It did not add supply, the profit model was unclear and it might require massive subsidies. He preferred the direction of Yidao Yongche, which added supply, but Yidao never brought prices down.
Later, Alibaba and Tencent each backed a company that continued burning cash; Didi expanded from taxis into express rides, premium cars and carpooling and completed a merger. All of that exceeded his understanding of China’s internet history at the time. “People can only earn money within the boundaries of their understanding.”
He is therefore not especially regretful. To invest in Didi, one had to enter extremely early and exit relatively early, or enter when the valuation was several billion dollars; later, the return was not necessarily attractive, let alone at a valuation of several tens of billions. His principle remains to act only with “70% confidence,” accepting that he will miss models with only a 20%-30% probability of success.
40. Invest in the Person or the Model? It Depends on Familiarity and Bet Size
屈田’s final answer is that early-stage investors can back people, but only when they know the team well and keep the check small. To make a large bet, “both the model and the people must work”; founder charisma alone is not enough.
Tuya Smart initially built a cloud photo album for phones. 屈田 worried that iPhone and Android would eventually add the native functionality, and they did. The team later pivoted to a smart-home cloud platform, showing that his early investment was not in the original plan but in long-term trust in the founders.
Waterdrop was also founded by people he knew. Both companies were valued at roughly RMB100M when first funded, providing a sufficient margin of safety. 屈田’s core judgment was: “Even if he doesn’t do this, he can succeed at something else.”
There are only 2 simple ways to raise certainty early: stay within ecosystems such as Alibaba, Tencent and J&T, observe executives over a long period and then invest; or wait until the operating data is clearer, accepting a higher valuation and lower return.
41. VC Competition Is Moving Toward Scale and Flexibility, but ATM Wants a Third Position
After more than 20 years watching China’s VC industry, 屈田 sees intensifying competition as the clearest change: funds compete on capital scale, young teams and brand. It is increasingly difficult for a fund head to win deals personally, and less room remains for new firms.
He agrees that the industry is developing a barbell structure: large-scale firms in the A16Z mold at one end and highly flexible small funds at the other. But greater scale usually compresses returns and requires enough people to manage it—neither is his preferred direction.
Globalization offers a path other than direct US-China competition. ATM does not need to fight Hillhouse, Sequoia, Boyu or Tencent for Chinese deals. It can fill the gap in overseas market judgment and ground capabilities where those firms are less familiar, turning former competitors into partners.
He has greater respect for firms such as DCM and Gaorong that do not compete purely on scale, noting that Gaorong’s early investment in Pinduoduo was not a function of fund size. The ultimate moat remains insight, with capital, ecosystem and people arranged around it: “Focus on the most important thing.”
42. Long Duration, Ease and “Fun” Make Up 屈田’s Final Investment Philosophy
Faced with hot deals, star investors and LP pressure for DPI, 屈田 acknowledges the industry’s brutality but refuses to be pulled by the current consensus. He distinguishes freedom from ease: freedom means not being constrained by the outside world, while ease means not being constrained by one’s own mind. Investing should be a vote for the objective truth one believes in.
Consumer businesses, digital currencies, AI and embodied intelligence take turns as hot themes; that does not mean companies outside the spotlight have no value. He uses an analogy to warn that valuing a company only by Wall Street standards is “like judging Marilyn Monroe only by her bust measurement.” Short-term price is not the same as total intrinsic value.
李翔 pointed out that LPs ultimately look at DPI. 屈田’s answer was that it is impossible to please everyone; he only needs to find long-duration capital willing to wait. After J&T’s listing, he said he “has not sold a single share.” If 3 to 5 more comparable investments emerge, he hopes to build an evergreen fund over 5 or 10 years.
Investing first attracted him because it combined economic returns, personal interest and the chance to change the world. Globalization is more interesting than changing China alone. He is no longer trying to become another 沈南鹏 or imitate any institution. He asks only whether a company can become a global champion—returning to the line that started it all: “Everything is for fun.”