On June 10th, Paul Graham, founder of Y Combinator and author of Hackers and Painters, gave a speech at Oxford University with a jaw-dropping title: “How to Earn a Billion Dollars.” I found his insights inspiring, so I’ve translated the key content. https://ruanyifeng.com/blog/2026/06/weekly-issue-401.html…

Tech Enthusiast Weekly (Issue 401): How to Earn a Billion Dollars - Ruan Yifeng’s Blog

Source: https://www.ruanyifeng.com/blog/2026/06/weekly-issue-401.html This newsletter records technology content worth sharing each week, published on Fridays.

This magazine is open source (https://github.com/ruanyf/weekly), and contributions are welcome (https://github.com/ruanyf/weekly/issues). There is also a “Who’s Hiring” (https://github.com/ruanyf/weekly/issues/10147) service that publishes programmer job listings. For cooperation, please contact via email (https://www.ruanyifeng.com/cdn-cgi/l/email-protection#235a4a45464d440d5156424d63444e424a4f0d404c4e)([email protected]).

Cover Image

In southern China, people wear sun visors made from giant lotus leaves. (via (https://www.sohu.com/a/906763935_121284943))

How to Earn a Billion Dollars

Paul Graham is the founder of Y Combinator, the largest startup accelerator in the United States, a recognized startup mentor, and the author of Hackers and Painters.

He has now retired, left Silicon Valley, and lives in the English countryside.

On June 10th this year, he was invited by a student society at Oxford University to give a speech (https://paulgraham.com/earn.html) with a title that was absolutely explosive: “How to Earn a Billion Dollars.”

I found the content quite interesting and have made some excerpts.

Some people seem to think that earning a billion dollars is impossible unless you use illegal or unethical means.

In my opinion, earning a billion dollars is indeed very difficult, but not impossible, and the opportunities are greater than most people imagine.

Twenty-one years ago in 2005, we founded the accelerator Y Combinator, and we have invested in about 6,500 startups to date.

So far, among the founders of these companies, about 30 have become billionaires. Moreover, many more are rapidly approaching this goal.

Out of 6,500 companies, assuming a total of 20,000 founders, with 30 of them earning a billion dollars, the chance is not that small.

The way they earned a billion dollars was by founding a successful company.

Recently, I was chatting with a founder. I asked about her company’s growth rate, and she said it was 93% last month. This means her net worth is likely growing at the same rate.

Let’s do a simple calculation.

Assume her net worth is currently $2 million, all invested in her own company. Then the company only needs to grow 500 times for her assets to reach a billion-dollar level.

How many months do you think it will take for her company to achieve 500x growth?

Assuming the company can maintain a monthly growth rate of 93%, then you simply calculate the logarithm of 500 with base 1.93, i.e., log(500, 1.93).

The answer is 9.45.

That is, starting from $2 million, maintaining a monthly growth rate of 93%, it only takes nine and a half months to achieve 500x growth, thus earning you a billion dollars.

Now you understand why, when I meet a founder, the first thing I ask is their growth rate.

You might say that a monthly growth rate of 93% is unrealistic. Then let’s change it to 15% monthly. After five years, how many times will you grow?

We calculate 1.15 to the 60th power (because five years is 60 months), 1.15^60 is approximately 4384.

This means that after five years, your company’s revenue will be 4384 times what it is today.

If the company’s current monthly revenue is $10,000, with this growth rate, the monthly revenue after five years will reach approximately $44 million, which is $526 million per year. By then, if you hold company shares like most founders, you will become a billionaire.

In fact, a 15% monthly growth rate corresponds to an annual growth rate of less than 5.5 times. Many startups can achieve or exceed this growth rate.

In short, if you start a company in your early twenties and maintain a high growth rate, becoming a billionaire by the time you’re thirty is absolutely possible.

The key to maintaining a high growth rate is that you must create products good enough that people recommend them by word of mouth, ensuring a steady stream of customers.

This is another reason I always ask founders about their growth rate first. The growth rate reflects whether they have made the right product.

Any project you genuinely believe is worth developing, no matter how absurd it sounds, is very likely to develop into a good startup idea.

No matter how absurd your project is, it cannot be more absurd than Justin.TV, which we invested in back in 2006.

That company had just one person: founder Justin Kan. He wore a camera on his head and walked around, live-streaming everything he experienced.

He later built a platform for others to stream like him. Ultimately, the company did quite well. You might have heard of it, but now it’s called Twitch.

The key to starting a business is to deeply understand a specific user group, so you can precisely build a product they truly want.

What do users really want? What can you do for them that significantly improves their lives?

This is entrepreneurial empathy, and it’s the quality we look for and cultivate in founders.

Earning a billion dollars through entrepreneurship ultimately comes down to two determining factors: growth rate and the duration of growth.

Growth rate depends on building a product that users love and are willing to share; the duration of growth depends on entering a huge market.

If your startup can grow at an exponential rate and capture a large market, its value will skyrocket, and as a shareholder, you will naturally become wealthy, likely earning a billion dollars.

How Speed Test Websites Make Money

In March this year, the outsourcing giant Accenture spent $1.2 billion (https://www.theverge.com/tech/889234/downdetector-ookla-speedtest-sold-accenture) to acquire Speedtest (https://www.speedtest.net/) and DownDetector (https://downdetector.com/).

Both are free websites — one for testing internet speed (pictured above), the other for checking if a website is down. This leaves many puzzled: why are free websites so valuable?

Insiders revealed (https://news.ycombinator.com/item?id=48339253) that the speed test website Speedtest is actually a very profitable business. Countless visitors from around the world use it daily to test their speed, and many software applications also access it to determine network conditions. As a result, it has a massive amount of real-world speed data from different regions and networks.

Speedtest makes money by selling this data. As the world’s largest test site, telecom operators are willing to purchase its data to improve their own networks. Each data set is quoted in six-figure dollars, with many customers purchasing annually, making Speedtest very profitable.

So, Accenture essentially acquired a data business: users generate the data, it organizes it, and sells it. Basically, free websites have only two ways to make money: advertising or selling user personal data or generated data.

Pull Requests Are Not Free

Richard Hipp, the author of SQLite, explained in an interview (https://www.youtube.com/watch?v=x8_ZZhRL3YU&t=1733s) (https://lobste.rs/s/aqk7vl/pull_requests_are_free_puppies) why his project never accepts external PRs.

Suppose you have a PR for SQLite. You come to me and say, “Hey, I developed a new feature for SQLite. Here’s my PR.”

You want me to merge it into the codebase, and you say, “Oh, it’s free.”

No, a PR is not free.

What you’re actually asking me is: you developed this great feature, and now you want me to maintain it for you, write documentation for it, test it for you, and maintain it for you for the next twenty-five years. That’s not free.

Linus once famously said: Free can mean free beer or free speech. But there’s another kind of free: free puppies. “Hey, I have a free puppy for you.” You know what I mean?

Submitting a pull request is like someone giving you a puppy. By the end of the day, you have a puppy in your house. You can’t throw it away — you have a moral obligation to take care of it until it dies a natural death.

I don’t want any free puppies.

One-Liner News

(1) A study found that grip strength (https://join1440.com/r/20950) (how firmly you can grasp an object) is a better predictor of mortality risk than blood pressure.

For every 5 kg decrease in adult grip strength, the risk of death increases by 16%.

(2) How expensive storage can be: Sandisk has launched a gaming-specific SSD for PS5 (https://games.gg/zh-CN/news/sandisk%E6%96%B0%E6%AC%BEps5-ssd%E5%94%AE%E4%BB%B7%E6%83%8A%E4%BA%BA%E8%B6%85%E8%BF%87%E4%B8%89%E5%8F%B0ps5-pro%E7%9A%84%E6%80%BB%E4%BB%B7/), with a capacity of only 8TB but a price tag of $3,000, three times the cost of a PS5 itself!

(3) Greece recently restored (https://apnews.com/article/greece-acropolis-restoration-parthenon-tourism-da06640fcd747498613d31b64dac369a) the famous Parthenon temple on the Acropolis of Athens, replacing missing marble pieces.

Below is what the temple looked like before the restoration.

Articles

1. A Detailed Explanation of the New HTTP QUERY Method (https://kreya.app/blog/new-http-query-method-explained/) (English)

HTTP has officially introduced the QUERY method in addition to GET and POST.

It is essentially a GET method with a body, allowing you to send a large number of parameters at once, which will not be cached by the server.

2. Anonymous Token Protocol PACT (https://www.cloudflare.com/press/press-releases/2026/cloudflare-collaborates-with-leading-browsers-to-develop-a-privacy-first-protocol-for-the-global-internet/) (English)

Cloudflare, together with three major browsers — Chrome, Firefox, Edge — announced the development of a new protocol to identify bot traffic.

The browser adds a token to HTTP requests from real users. The server uses this token to determine whether the visitor is a bot. Details are still unclear. You can refer to Mozilla’s introductory article (https://hacks.mozilla.org/2026/06/pact-anonymous-credentials-for-the-web/); in theory, this would eliminate the need for CAPTCHA verification.

3. The Rise and Fall of OpenCL (https://www.modular.com/blog/democratizing-ai-compute-part-5-what-about-cuda-c-alternatives) (English)

A retrospective article. In the early 2000s, several companies began envisioning a universal C++ framework for GPU operations. However, due to conflicting interests among parties, reaching consensus in the project committee proved difficult, and ultimately it failed.

In the end, NVIDIA’s CUDA framework replaced it and became the standard method for GPU operations, making NVIDIA the dominant player in AI hardware.

4. Stop Using JWT (https://gist.github.com/samsch/0d1f3d3b4745d778f78b230cf6061452) (English)

This article argues that JWT tokens should not be used to maintain user login state; that is the function of cookies. The only suitable use case for JWT is transferring user state from one machine to another.

5. I Stored a Website in a Favicon (https://www.timwehrle.de/blog/i-stored-a-website-in-a-favicon/) (English)

This article presents an interesting trick: the actual content of the webpage is hidden inside the favicon file, and then decoded and rendered using JavaScript.

Tools

1. Lore (https://github.com/EpicGames/lore)

A version management system open-sourced by the game company EpicGames. Compared to Git, its main feature is version management for binary files.

It splits large binary files into chunks for storage. Each commit only saves the chunks that have changed.

2. DNS Pick (https://github.com/palemoky/dnspick)

A command-line DNS optimization tool that combines average latency and resolution success rate to select the best DNS server balancing speed and stability. (Contributed by @palemoky (https://github.com/ruanyf/weekly/issues/10311))

3. GitFolio (https://github.com/azhai/gitfolio)

A lightweight Git repository management system, similar to Gitea, supporting syncing repository data from GitHub mirrors. (Contributed by @azhai (https://github.com/ruanyf/weekly/issues/10316))

4. ssh-at (https://github.com/baerwang/ssh-at)

A GUI management tool for ~/.ssh/config. (Contributed by @baerwang (https://github.com/ruanyf/weekly/issues/10330))

5. LockIME (https://github.com/oomol-lab/LockIME)

An input method locking tool for macOS that can specify the default input method for different applications. (Contributed by @BlackHole1 (https://github.com/ruanyf/weekly/issues/10279))

6. Cover Maker (https://github.com/eternityspring/article-tools)

A web-based tool for creating cover images. (Contributed by @Hao4Wang (https://github.com/ruanyf/weekly/issues/10276))

7. PowerLens (https://github.com/luyangkk/powerlens)

An Oh-My-Zsh plugin that displays real-time power consumption, battery status, CPU, CPU temperature, fan speed, memory, and network traffic in the command prompt. (Contributed by @luyangkk (https://github.com/ruanyf/weekly/issues/10345))

8. MyKVM (https://github.com/XxMinor/mykvm)

A cross-platform KVM software that allows macOS, Windows, and Linux to share a single set of keyboard, mouse, and clipboard on the same local network. (Contributed by @fc221 (https://github.com/ruanyf/weekly/issues/10373))

9. ai_caption_video (https://github.com/alexchan197611/ai_caption_video)

An open-source Windows application for generating Chinese short videos with large captions, supporting keyword highlighting, subtitle animation, local TTS dubbing, and voice cloning. (Contributed by @alexchan197611 (https://github.com/ruanyf/weekly/issues/10378))

10. AnyDrag (https://github.com/XueshiQiao/AnyDrag)

A macOS utility that allows dragging, resizing, maximizing, and tiling windows without needing to hold the title bar. (Contributed by @XueshiQiao (https://github.com/ruanyf/weekly/issues/10398))

11. Direct Light (https://github.com/oukeming64-tech/direct-light)

A web-based simulation of studio lighting. (Contributed by @oukeming64-tech (https://github.com/ruanyf/weekly/issues/10404))

12. JSOS (https://jsos.dev/)

Based on Webcontainer technology, runs Node.js applications in the browser. Data and code are stored locally. (Contributed by @jsos-dev (https://github.com/ruanyf/weekly/issues/10410))

AI-Related

1. Fishword (https://github.com/Chenggou1/fishword)

A plugin for the programming agent Pi. While waiting for the AI to generate code, it displays a window for learning vocabulary. (Contributed by @Chenggou1 (https://github.com/ruanyf/weekly/issues/10338))

2. OnePagent (https://github.com/sligter/OnePagent)

An open-source, browser-native single-file AI agent workbench. Just open an HTML file to perform AI operations. (Contributed by @sligter (https://github.com/ruanyf/weekly/issues/10363))

3. SlopGuard (https://github.com/Blue-B/slopguard)

A GitHub app that automatically scores PRs and issues for quality, used to filter out low-quality submissions. (Contributed by @Blue-B (https://github.com/ruanyf/weekly/issues/10275))

4. TiyGate (https://github.com/tiylabs/tiygate)

A self-hosted AI gateway that can automatically switch between multiple subscription plans and provides a unified management panel. (Contributed by @jorben (https://github.com/ruanyf/weekly/issues/10401))

5. Mediary Scout (https://github.com/fancydirty/mediary-scout)

An open-source agent for searching TV shows across various resources and saving them to your cloud drive. Requires self-deployment. (Contributed by @fancydirty (https://github.com/ruanyf/weekly/issues/10412))

Resources

1. Cutting Firewood (https://screen.toys/firewood/)

A web mini-game that is exceptionally realistic.

2. Solar Wanderer (https://github.com/hyqzz/Solar-Wanderer)

A browser-based realistic scale model of the solar system, including 8 planets, the Moon, and 21 satellites. (Contributed by @hyqzz (https://github.com/ruanyf/weekly/issues/10349))

3. PTP Time Synchronization Technology Book (https://github.com/Lularible/ptp-book/tree/main/chapters)

An open-source technical book that introduces the PTP/IEEE 1588 precision time protocol in plain language and analyzes the LinuxPTP source code. (Contributed by @Lularible (https://github.com/ruanyf/weekly/issues/10284))

4. Cosmodial (https://killedbyapixel.github.io/Cosmodial/)

A web-based starry sky simulator to explore the universe in your browser.

Images

1. Virtual Map (https://www.jerrysmap.com/the-map)

An American artist has a habit of doodling in his free time. Whenever he has a moment, he randomly draws color blocks on paper.

One day, he suddenly realized that these doodles, when put together, looked like a virtual map.

So he began to seriously devote himself to this project, piecing together over 4,000 doodles into a map of a virtual world.

All the color blocks combined form a perfect circle for this virtual world.

Digest

1. The History of the Meter (https://www.abc.net.au/news/science/2025-05-20/metre-treaty-anniversary-metric-system-measurement-metrology/105302024)

At the end of the 18th century, the French Revolution broke out.

At that time, France had countless measurement systems, creating utter chaos. There was even one unit of length set by King Henry I of England in the 11th century: one yard was the distance from the king’s nose to the tip of his outstretched arm.

The French revolutionaries decided to create a completely new, nationally unified system of measurement. The new unit of length was called the “meter,” defined as one ten-millionth of the distance from the North Pole, through the Paris Observatory, to the Equator.

But no one knew exactly what that one ten-millionth was.

The task of calculating it fell to two astronomers. Seven years later, in 1799, they submitted their final measurement result to the French Academy of Sciences — this is the length of one “meter” as we know it today.

The Academy then had this length made into platinum bars. A total of 30 bars were produced and distributed across the country to make people aware of the new measurement standard “meter.” These platinum bars were called “meter prototypes.”

Modern instruments measuring a preserved platinum bar found it to be quite accurate, only 0.2 mm shorter than the modern standard.

This definition of the meter was used until the second half of the 20th century. At that time, scientists proposed using the wavelength of light to measure distance, since the wavelength of light is constant. Thus, the definition of the meter was revised: 1 meter was equal to 1,650,763.73 times the wavelength of the red-orange light emitted by krypton atoms when an electric current passes through a lamp filled with krypton gas.

However, this red-orange wavelength was inconvenient for measuring extremely small distances. In 1983, the definition of the meter was changed again to use the speed of light: 1 meter is the distance light travels in a vacuum in 1/299,792,458 of a second.

Quotations

Improve yourself in ways that benefit others. This is what we’re after.

-- Why We Hire Junior Engineers (https://newsletter.kentbeck.com/p/hey-n00b-we-didnt-hire-you-to-complete)

The European standard for a livable residence is that you can see at least three trees from inside, the tree coverage rate in the neighborhood is at least 30%, and there is a park within no more than 300 meters.

-- Can You See Three Trees? (https://www.not-ship.com/can-you-see-three-trees/)

When I read books published before 2022, I know that every word was manually input, manually proofread, manually edited, and manually proofread again. Somehow, this affects me and makes me value the book and its content more.

-- Books Before 2022 (https://notes.lorenzogravina.com/musings/pre-2022-books)

I found myself trapped in the situation of many engineers at large companies. My title and salary were only slightly higher than a junior engineer’s, but the work I did every day was at the “senior” or “staff” engineer level.

Every time I tried to get promoted, I failed. I felt like I was constantly solving problems two levels above my own job title, and the only reward I got was more work assigned by my superiors.

-- Why I Left YouTube (https://zhach.news/how-i-left-youtube/)

Looking Back

What Does an 8000mAh Phone Battery Tell Us? (https://www.ruanyifeng.com/blog/2025/06/weekly-issue-354.html) (#354)

The Most Popular Color (https://www.ruanyifeng.com/blog/2024/06/weekly-issue-304.html) (#304)

Life is a Long-Board Problem (https://www.ruanyifeng.com/blog/2023/05/weekly-issue-254.html) (#254)

How to Survive a Pandemic, Layoffs, and War (https://www.ruanyifeng.com/blog/2022/04/weekly-issue-204.html) (#204)

(End)