Robin argues that Taiwan’s digital and software startups lack a process of branching out. PayPal has generated a large network of startups and investments, potentially spanning thousands of companies and dozens or hundreds of unicorns. Taiwan’s hardware sector has similar networks: Acer gave rise to BenQ, Wistron, ASUS, and others, while ITRI generated many listed companies. Yet Trend Micro, founded in 1988, remains Taiwan’s most influential software company without producing a comparable cluster. 91APP and Appier, described here as having listed only four or five years earlier, have not yet generated enough surrounding startups. Taiwan needs companies or accelerators such as PayPal, Palantir, or YC that continually give rise to new ventures.
Key points
PayPal, Acer, and ITRI have all generated branching networks of companies.
Taiwanese software companies have not yet produced large startup clusters.
The ecosystem needs companies or accelerators that generate further startups.
Excerpts from Wu Jun’s letters plus Robin’s commentary (as of October 2024)
Reading guide
The opening excerpts from Wu Jun warn against abandoning your strengths to chase trends. They cite Buffett on keeping well-performing holdings and Eric Schmidt on reinvesting in a rapidly growing core business. Robin adds that Taiwanese startups often rush to open second and third fronts before securing the first. They may still be overtaken or replaced. If the first market offers substantial room and the startup’s share remains low, establish a firm footing before taking another step.
Key points
Do not rush into new fields while the core business is thriving.
Opening a second front too early can leave the first vulnerable.
With low market share, first establish a firm footing.
Robin argues that cash-flow management matters to every company. Large businesses cutting staff after overhiring before the pandemic are considering cash flow as well as focus and efficiency. For startups, running out of cash often means the end. But managing cash does not mean refusing to spend. Sometimes frugality is essential; sometimes a rare market opening requires expansion or even a calculated gamble. A startup that plays only safe cards is unlikely to overtake incumbents. Founders need exceptional insight and judgment about timing—to create miracles rather than wait for them.
Key points
A break in cash flow often means the end of a startup.
The issue is when to conserve and when to expand, rather than never spending.
Startups that play only safe cards rarely overtake incumbents.
Robin offers eleven startup-mentoring observations. A need ranked thirteenth by customers is a nice-to-have, even if they acknowledge it exists; investors prefer must-haves. Before committing resources, conduct open-ended interviews with thirty target users, one-on-one for two hours each. Test willingness to pay: actual customer payments are the strongest signal. Treat free pilots cautiously, since passing a trial does not guarantee payment. Recycling ventures need full cost-benefit analysis because customers may not pay higher costs. Avoid projects requiring expensive market education where possible. Interview before building even an MVP, and heavily discount stated willingness to pay.
Key points
Establish whether demand is must-have or merely nice-to-have.
Conduct open-ended interviews with roughly thirty target users.
Actual customer payments provide the most convincing validation.
Excerpts from a Dedao audio summary of 張忠謀傳 plus Robin’s commentary (as of May 2025)
Reading guide
The opening audio-book summary describes TSMC targeting European and US markets from its first day and growing with customers such as Qualcomm, Broadcom, NVIDIA, Apple, and AMD. Robin adds that earning customers’ trust requires doing certain things better than they can themselves. Only then will they share key technologies and develop the next generation together. TSMC’s achievement, he argues, comes from moving alongside a group of world-leading companies, rather than developing in isolation and assuming it can beat everyone upon emerging.
Key points
TSMC grows together with its major customers.
Doing things better than customers can themselves builds trust.
Advance alongside leading companies rather than developing in isolation.
Excerpts from a Dedao audio summary of 馬斯克邏輯 plus Robin’s commentary (as of December 2025)
Reading guide
The opening audio summary of 馬斯克邏輯 contrasts Errol Musk with his son: similar personalities did not produce similar achievements, reflecting differences in timing and environment. Robin adds that Taiwan has its own entrepreneurial setting, positioned between the United States and China, near Japan and rapidly growing Southeast Asia. People who grew up in Taiwan and choose to start companies there should consider how to use its geographical advantages and resources.
Key points
Individual ability also needs favorable timing, place, and people.
Taiwan sits between the United States and China, two major innovation powers.
Use Taiwan’s geographical advantages and resources.
Robin’s commentary with a reprint from the 低風險創業 series in the middle
Reading guide
Robin argues that entrepreneurship is not an exam. A tutor can summarize material and teach test-taking techniques, but a few startup classes or advisers cannot substitute for learning through practice. The middle reprint describes success stories as rough frameworks and urges founders to find their own path rather than copy appearances, using restaurant brands as examples. Robin concludes that even Jack Ma or Pony Ma could not reproduce Alibaba or Tencent in a different time and setting. Details change, and some teachers and advisers have no actual entrepreneurial experience.
Key points
Entrepreneurship has no master teacher; do not place blind faith in classes.
Success cannot simply be copied across different times and circumstances.
Robin argues that a startup building at the application layer—on a technology giant’s platform—may lack strong barriers and differentiation. A better way to win is exceptional product quality and user experience. A competent team can produce a 60-point product, but a giant can casually do the same; matching it means losing. Differences emerge between 60 and 80, while winning truly requires 90 to 100, with each additional half-point to point extremely difficult. Use your product and competitors’ products as a user would, examining differences throughout the workflow and whether they actually solve users’ problems.
Key points
Application-layer startups with weak barriers need to win through product and user experience.
A product scoring the same 60 points as a giant’s will lose.
Winning requires 90 to 100 points, and every additional point is difficult.
Robin’s commentary with a reprinted story; the passage names Liu Run and Dedao Headlines
Reading guide
Robin urges platform founders to distinguish traffic brought by influencers, key opinion leaders, or teachers from traffic created by the platform. A star’s traffic can disappear as quickly as it arrives. He compares this with artist-management relationships: recording companies may continue selling recordings through their copyrights, while management contracts simply end. One early-stage approach is to offer favorable terms to attract top stars, quickly build the platform’s name, then distribute traffic to second- and third-tier creators and newcomers. Distinctiveness and a moat can create a platform flywheel. Without its own brand-driven traffic, a platform can be dominated by stars who move between platforms for better terms.
Key points
Distinguish traffic owned by the platform from traffic belonging to its stars.
Attract top stars early, then distribute traffic to smaller creators and newcomers.
Build brand-driven traffic and a moat to create a platform flywheel.
Robin’s commentary following excerpts from Wan Weigang’s 精英日課
Reading guide
The opening story about sociologist Corey Keyes illustrates how environment shapes people. Robin argues that Taiwan’s best may rank only in the middle in Asia and need enormous effort to keep up globally. He compares moving from Taiwan’s professional baseball league to Japan and Major League Baseball, and describes a software team of over 200 facing global competition despite leading Taiwan. Some ventures, such as semiconductors, can start in Taiwan because it has world-leading talent, customers, and technology. In fields where Taiwan has little global presence, founders should carefully consider targeting the world or their main market from the outset.
Key points
Each step to a higher level brings a very different challenge.
In fields where Taiwan leads, such as semiconductors, startups can grow alongside its major companies.
Where Taiwan lacks an advantage, consider targeting global markets from the outset.
Robin’s commentary following excerpts from 創新工場投資人的投資創業課
Reading guide
Continuing his discussion of challenging industry consensus, Robin argues that the next technology giant will necessarily do things differently from today’s giants. To win, you cannot follow the existing winners’ rules; within their rules, you will always lose. Rather than hoping to do better, create something different. The world may not need a second Microsoft or Google, but many areas still need innovation and breakthroughs.
Key points
Do not follow the existing winners’ rules.
Rather than doing better, create something different.
Many areas of the world still need innovation and breakthroughs.
Robin’s commentary (as of December 2024), preceded by excerpts attributed to 中國企業家
Reading guide
Robin notes that most people were skeptical when Lei Jun announced Xiaomi would make cars, yet the SU7 sold over 120,000 in its first year. What impresses him most is Lei’s ability to catch up after entering late. Dozens of phone makers preceded Xiaomi, and electric vehicles also appeared to be a crowded market, but the team’s exceptional execution let it catch up. Other lessons—changing one’s thinking, setting aside pride, and learning from competitors—are offered for reference.
Key points
Latecomers can still catch up and take the lead.
In a crowded market, execution determines whether a latecomer can catch up.
Changing one’s thinking and learning from competitors offer lessons to consider.
Reprinted FDE article and Robin’s commentary; original author not identified
Reading guide
This reprinted article on forward-deployed engineering discusses how B2B startups choose reference customers. Such customers provide revenue and peer endorsement, while field experience shapes product direction. Well-funded requests that build no reusable capabilities can consume the entire team. Robin adds that knowing which projects and customers to decline matters as much as knowing what to pursue. The Harvey and Palantir examples are relayed from the article and were not individually verified in this review. Their figures should not be read as achievements independently confirmed by this site.
Key points
Reference customers matter both for their endorsement and their product feedback.
Distinguish reusable experience from custom requests that exhaust resources.
Robin: choose which customers not to serve; not every available dollar is worth earning.
Notes from Dedao’s audio summary of 精進, with Robin’s commentary
Reading guide
This summary of Dedao’s discussion of 精進 warns against preparing forever without finishing. A minimal product need not be a complete company or complex software; it can be a structured account of experience that others can use. Deliver something first, then improve it through feedback. Robin adds: learn and adjust by doing, rather than making unreadiness an excuse. The Thiel program described uses its early under-20 design; later rules differ, as explained in the source note. This is not encouragement to leave college and start a company without assessing the risks.
Key points
First complete a small, usable output that carries your ideas.
Improve through feedback instead of preparing indefinitely.
Robin: learn and adjust by doing; do not let unreadiness block action.
Reprinted article on AI-powered one-person companies, with Robin’s brief commentary; original author not identified
Reading guide
A reprinted article on AI-powered one-person companies, followed by Robin’s brief commentary; the original author is not identified. Before starting, recognize the gap between observing and doing, and the reality of not yet knowing how to run a company. Three layers matter: distinguish resources supplied by an organization from personal capabilities; build a working business system from positioning through customer relationships; and use AI as a long-term collaborator that amplifies judgment and capacity. AI helps only when you enter the market and improve your judgment through experience. Robin favors leaner teams and more careful hiring.
Key points
Distinguish an organization’s resources from your own capabilities.
Make positioning, products, customer acquisition, sales, and customer relationships work together.
Robin favors leaner teams and greater care in hiring.
These notes tagged 未來回推法 describe leaders’ need to manage today while looking further ahead. Three horizons cover optimizing the core business, cultivating new growth opportunities, and working backward from a vision five to ten years ahead to today’s strategy. The central proposition is that the future is designed rather than predicted, and strategy reshapes the future rather than merely extending the present. Robin adds: begin with the end in mind, but take each present step well.
Key points
Optimize the core business while cultivating new opportunities.
Reconsider today’s choices from a long-term vision.
Robin: begin with the end in mind without neglecting each present step.
Reprinted FDE article; original author not identified, with no separate commentary by Robin
Reading guide
This reprinted FDE article, whose author is not identified, describes trust in enterprise procurement: vendor publicity ranks below analyst reports, public peer cases, and ultimately private peer recommendations. FDE can generate material customers share privately. Three actions matter: turn delivery results into a story with numbers, people, and contrast, preparing a one-page summary, three graphics, and a thirty-second version; make customers’ business success part of their own professional standing, illustrated by Walgreens and J.D. Power; and manage endorsement backlash, because reference-customer failures can spread faster than successes.
Key points
The excerpt argues that enterprise buyers value private peer recommendations.
Present delivery results through specific numbers, people, and contrasts.
Reference-customer failures can also spread quickly; manage delivery risk.
This 2023 column distinguishes an ordinary business from a venture-backed startup through growth speed, scalability, and capital needs. First clarify what you want to build, the life you want, and the sacrifices you are willing to make. Then decide whether fundraising and scaling are the right path.
Key points
First determine whether your business needs rapid growth and scale.
VC return structures shape investment choices and do not suit every good business.
Begin with the end in mind and choose a path aligned with personal goals.
Investors assess a market’s ceiling, but founders should not simply chase media excitement. This column brings together industry size, niche needs, and personal passion. Begin with a problem you genuinely understand and want to work on for years, rather than choosing a venture for fundraising.
Key points
Market size shapes growth potential, but popularity does not make entry suitable.
Solve a small group’s real problems first, then seek opportunities to expand.
Build for your own ideals rather than substituting investors’ expectations for direction.
Facebook’s early investment story illustrates the value of trusted introductions: trust, industry understanding, and an introducer willing to share the risk. The author then compares fundraising to concentric circles, reminding founders to value all money equally and accept the responsibility it brings, whoever provides it.
Key points
Trusted introductions help early teams cross the trust barrier.
An introducer’s industry knowledge and willingness to co-invest influence the result.
The first investment also brings new responsibilities.
Before investing in a team and product, establish who needs it, why the need is urgent, and whether they will pay. Starting with the technology-first approach common in academic ventures, this column emphasizes priorities, niche customers, and continual validation rather than attributing every sales problem to insufficient market education.
Key points
First identify the target customer’s most urgent problem.
Clarify needs through experiments and validation, not confidence in technology alone.
Start with a small, clearly defined problem and win paying customers.
Using Threads’ early development alongside Tesla, SpaceX, and Amazon, this 2024 column argues for focusing on customer problems and distinctive solutions rather than merely catching incumbents. The examples reflect observations at the time of writing, not conclusions about the products’ later development.
Shareholders are long-term partners, not funding sources you can replace at will. Comparing CVC and VC resources, interests, and risks, this column advises founders to investigate investors’ reputations while cash is still sufficient and consider whether the shareholder mix supports long-term cooperation.
Key points
Equity partnerships last and are difficult to exit; choose shareholders carefully.
Strategic resources can bring labeling, conflicts of interest, and disclosure risks.
Fundraising is mutual selection; investigate investors’ collaboration records.
Drawing on FITI and research-commercialization mentoring, the author supports government bearing early technology risk while urging academic teams to return to customer needs, mission, and commitment. Interview methods, peer support, and willingness to pay move ventures forward more effectively than technology demonstrations alone.
Key points
Government and VC play different roles at different innovation stages.
Explore customer problems through open interviews rather than leading questions.
Beyond technology, clarify motivation, team commitment, and demand.
Looking back at crises, this column prepares startups for a funding winter: prioritize customer revenue, concentrate on the core, and offer investors a new plan worth backing. It is a strategic view from 2025, not a real-time forecast of 2026 fundraising.
Key points
Strengthen survival through paying customers and the core business.
Help yourself before fundraising; abandon expansion that does not support core value.
Investors need a viable plan, not merely a request for winter provisions.
Commentary notes; the original post does not identify a separate source
Reading guide
In this note from the end of 2020, Robin suggests that continually adding features may reveal uncertainty about whom a product serves and what it solves. Rather than building an all-encompassing platform, make the first version simple, direct, and relevant to target customers. This is a management perspective, not a causal claim that every complex product must fail.
Key points
Clarify target customers and the problem they will pay to solve.
Extra features cannot replace confidence in the core need.
Concentrate limited resources on the most important breakthrough.
Commentary notes; the original post does not identify a separate source
Reading guide
This note treats strategy as trade-offs. Knowing what not to do creates depth in what matters most. Using quantity, speed, quality, and economy, Robin warns against confusing rapid hiring or broad activity with irreplaceable expertise. Developing one important capability deeply can be stronger than following everywhere. Claims about how many dimensions produce victory or defeat are rhetoric about trade-offs, not a quantitative success formula.
Key points
Strategy includes clearly stating what not to do.
Team growth and breadth do not equal depth of capability.
Management notes followed by Robin’s commentary; the opening does not identify a separate source
Reading guide
Robin cautions product teams against merely copying competitors or letting one large customer turn a reusable product into bespoke project work. Preserve your own industry understanding and make products easy to adopt and scale. The note allows an exception for specialized products with long implementations: fewer customers paying higher prices can be a viable model, rather than rejecting such products outright.
Key points
Competitive intelligence cannot replace your own industry judgment.
Distinguish general product needs from customization for one customer.
Consider implementation complexity alongside customers, pricing, and the scaling model.
Commentary notes; the original post does not identify a separate source
Reading guide
This note reverses the planning sequence: describe the desired future, assess today’s position, then gather the people, money, and resources needed. Robin’s stone-soup metaphor describes resource integration and contrasts with beginning from what is already available. This is a goal-oriented perspective, not a claim that startups can know outcomes in advance or must exclude experimentation.
Key points
Describe the goal before assessing present conditions.
Fill resource gaps through people, funding, and cooperation.
Goal orientation and starting with resources at hand are different perspectives.
More accessible funding, abundant incubators and accelerators, founder-friendlier terms, and AI have lowered entry barriers. Success is harder: abundant capital fuels irrational, unprofitable competition; talent costs more and is harder to retain; accelerating technology makes investors hesitate. Understanding the rules and everyone’s motives still offers a chance to win.
Key points
Funding, incubators, terms, and AI make starting easier.
Unprofitable growth spreads through competition, while rapid imitation shrinks the window.
Low birth rates force startups to compete directly with technology giants for talent.
Rapid responses by AI giants can make customers and investors hesitate.
Robin begins by arguing that disruptors look different and achievement requires differentiation. He uses analogies: Nokia’s feature-phone dominance was challenged by Apple’s smartphone market, not a stronger feature phone; a Baidu challenger might enter like Toutiao rather than build a better search engine. He similarly imagines TikTok rather than another social website challenging Facebook—an analogy, not an established outcome. A Starbucks challenger might open a delivery-and-takeaway online market like Luckin rather than copy its model.
Key points
A disruptor will look completely different.
Nokia’s challenge came from smartphones, not better feature phones.
TikTok versus Facebook is the author’s analogy, not an established outcome.
A Starbucks challenger could develop a different delivery-and-takeaway market.
Rather than fighting in a crowded market, define a clear audience, enter through a small opening, dominate a segment, then expand. The clearer the target customer’s profile, the easier it is to establish a place in their mind.
Key points
Find a clear target audience rather than fight in a crowded market.
Start with a narrow opening, dominate the segment, then expand.
A clearer customer profile strengthens positioning in customers’ minds.
Described as a ‘little Tencent of the Middle East,’ Yalla began in Dubai with voice-based social networking and listed on the NYSE in September 2020. It illustrates building for the target market from the outset. Founder 楊濤 spent years with ZTE in the UAE and understood local needs.
Key points
Described as a “little Tencent of the Middle East,” Yalla began in Dubai with voice-based social networking and listed on the NYSE in September 2020.
It illustrates designing for the target market from day one.
The founder’s ZTE experience in the UAE supported local understanding.
Do not abandon customer needs and product goals to please investors.
Excerpt from 范冰’s 前線部署工程師 with Robin’s commentary; the original post did not name the author
Reading guide
Before writing the first line of code at an enterprise customer, map data, processes, organization, systems, and politics. Two weeks of diligence may look like pure cost but can prevent six months of work in the wrong direction. Robin argues that enterprise projects often fail because the team has not understood internal organization, responsibilities, and interests.
Key points
Map the customer’s situation before writing the first line of code.
Prepare five maps: data, processes, organization, systems, and politics.
Two weeks of diligence can prevent six months of misdirected investment.
Robin: understand the customer’s organization, responsibilities, and interests.
Before expanding overseas, ask whether the market actually wants your project. In Singapore, for example, investment may follow naturally when a project is both a globally recognized Taiwanese strength—such as semiconductors, AI hardware, IC design, or critical components—and something Singapore wants to develop. If neither condition holds, the effort may yield little. The central questions are whether you can prove world-class strength and how badly the market needs you.
Key points
Before expanding overseas, ask whether the market wants your project.
Investment may follow naturally when the project matches both Taiwan’s recognized strengths and the host market’s priorities.
If neither condition holds, the effort may yield little.
Prove that you are world-class and that the market strongly needs you.
Startup teams often rush to expand across regions and countries without a team that can consolidate and protect their gains. Expansion becomes meaningful only after the company has established a solid home base.
Key points
Expansion matters only when the company can retain users externally and has an internal team to support them.
A team that opens new territory must be matched by one that cultivates and protects it.
Establish a solid home base before expanding outward.
Opening section reprinted (original author not identified); closing commentary by Robin
Reading guide
Test startup ideas on a small scale and at low cost, changing one variable at a time. A street stall offers low costs and rapid feedback. Robin considers it a more grounded starting point than imagining customers behind closed doors.
Key points
Street stalls offer low costs and fast feedback.
Change only one variable at a time.
Robin: direct experience is more grounded than imagining customers behind closed doors.
Excerpt from 范冰’s 前線部署工程師, with Robin’s commentary (the original post did not name the author)
Reading guide
Enterprise AI delivery can draw on three kinds of partners: cloud providers and large platforms, consultancies and systems integrators, and customers’ own customers and supply chains. Robin adds that potential partners in Taiwan include the Big Four accounting firms, telecom operators, and systems integrators.
Key points
Three partner groups can support enterprise AI delivery: cloud providers and platforms, consultancies and systems integrators, and customers’ customers and supply chains.
Customers’ committed cloud spend is a practical lever for co-selling.
Partners can provide customer relationships that would take three years to build yourself.
Robin: potential partners in Taiwan include the Big Four accounting firms, telecom operators, and systems integrators.
Excerpt from 范冰’s 前線部署工程師, with Robin’s commentary (the original post did not name the author)
Reading guide
High-caliber FDE delivery capacity is always scarce, making scheduling a strategic tool. Establish a profitability threshold, prioritize strategic value rather than contract size, turn the waiting period into a service, and never promise more concurrent projects than capacity permits. Robin recommends dividing vague large contracts into smaller projects that can be delivered and validated quickly.
Key points
Scarce FDE delivery capacity makes scheduling itself strategic.
Set a profitability threshold: small contracts cannot support expensive FDE staffing.
Prioritize strategic value over contract size and make waiting time productive.
Robin: divide vague large orders into smaller, quickly validated projects.
Excerpt from 范冰’s 前線部署工程師, with Robin’s commentary (the original post did not name the author)
Reading guide
Most technical proposals describe features rather than business value. A strong proposal has four layers: business outcomes, a path for validating value, an implementation plan, and risks with responses. In an enterprise market damaged by overpromising, willingness to be judged by business results is a valuable sign of sincerity.
Key points
Most technical proposals discuss features instead of business value.
Use four layers: business outcomes, value validation, implementation, and risk responses.
Willingness to submit to business-result verification signals sincerity in enterprise markets.
