# How Chess.com Became the World’s Biggest Chess Community with CEO Erik Allebest
<!--
ref: @content/youtube/no-priors/2026-08-13-the-evolution-of-chess-in-the-age-of-ai-with-chesscom-ceo-er
version: 3
updated_at: 2026-10-31T19:18:38.783Z
-->

Erik Allebest (No Priors, 2026-08-13): Allebest says Chess.com grew big because the team stayed focused on the player experience, paid for growth out of revenue, and changed what it means to call yourself a chess player.

**[Watch on YouTube](https://www.youtube.com/watch?v=hplb9IHWuw0)**


*Synthesized from Erik Allebest on No Priors, 2026-08-13.*

Erik Allebest co-founded Chess.com and is its CEO. He and a friend bought the domain out of a bankruptcy auction in 2005, and he grew it without venture money into the largest chess platform in the world. The No Priors host interviews him in this episode.

## Core Thesis
Allebest says Chess.com grew big because the team stayed focused on the player experience, paid for growth out of revenue, and changed what it means to call yourself a chess player. He sees chess as the longest-running test of what people do once machines beat them at something: 30 years after Deep Blue, more people care about chess than ever. In his view, people want to build human skill and watch other humans perform even when a computer can do better. AI helps players learn faster, but they still have to put in the practice.

## Key Insights
- **Scale today:** about 10 million daily active users and 40–50 million monthly active users (summer runs lower). There are more than 250 million registered members. Revenue should come in a little over $200 million this year. The team is about 650 people, all remote. The company's mission has three parts: make members happy, grow the game, and be a great place to work.
- **Origin:** in 2005 Allebest and a friend paid $56,000 for the chess.com domain at a bankruptcy auction. The host's intro says about $55K. The previous owner had been building chess software and couldn't pay its debts. The first goal was a chess community "like MySpace."
- **Turned down by VCs:** Allebest was at Stanford Business School and pitched Sand Hill Road investors. Most of them told him the idea couldn't be funded and he should get a real job. Classmates wouldn't join either.
- **Growing on cash:** Chess.com launched in 2007, right as he graduated. Within about 18 months it charged memberships for online chess learning, and it became profitable soon after. Funding came from proceeds of earlier businesses he had sold, plus a small loan from his co-founder's mom that was later repaid. They hired each new role (community manager, developer, sysadmin, designers) only once revenue covered it. They kept saying they'd never pass 50 people, then 100, and stopped making the prediction.
- **Doing the opposite of the standard playbook:** the advice back then was to recruit a Stanford technical co-founder, raise a big round, rent an office, chase a big market, and spend on customer acquisition. Allebest teamed up with a friend from San Jose State, worked remotely, raised nothing, spent $0 on paid acquisition, and picked a tiny market.
- **Competition:** the earlier field was FICS (Free Internet Chess Server), ICC (Internet Chess Club) and Yahoo Chess. Chess.com won by being free to play, running in the browser, and feeling good to use. Chess can't be patented, and the app stores hold thousands or tens of thousands of chess apps, so Allebest says product, community and content decide who wins. He describes Lichess as the open-source "Linux" of chess and says the two coexist.
- **When it was obviously a big business: 2024.** Before COVID and *The Queen's Gambit* the site had about 1 million DAU. The 2020–2022 spike could have been a fad "like treadmills and sourdough bread," but usage settled only a little below the 2022 peak. In 2023 a second wave hit, driven by short-form video, the Mittens bot, the cheating scandal and kids playing in schools. After that wave the baseline stayed higher again. That was when the team accepted that growth was compounding. Allebest says the growth was "white knuckle," with scaling problems across product and tech, and that only the past year has felt calm.
- **Changing who counts as a chess player:** players used to be taken seriously only around a 2000 rating and above. Chess.com celebrates mistakes like hanging your queen, and treats someone rated 500 as a real player. Allebest cites Louis Vuitton fashion shoots built around chess and Lewis Hamilton playing on his phone in his jet. The host says about 20% of kids in her town do after-school chess.
- **Ownership history:** the company is still bootstrapped, and no primary capital has ever gone in. Early angels who loved chess bought small stakes, promised to hold forever, and sold a few years later. Allebest disliked most of the private equity firms he met. General Atlantic, around 2020–2021, was the first PE firm that bought into the mission-first approach, and it bought shares secondary from existing investors. At General Atlantic's own fundraising cycle, CVC came in, chosen for its experience in gaming, sports and community businesses. General Atlantic reinvested alongside CVC.
- **PE made the company better, against the stereotype:** the investors shared industry knowledge, got into product details, and pushed for better forecasting, reporting and operations. Allebest says better reporting improved how the company runs, beyond what it gives investors.
- **How engines changed chess:** after Deep Blue, many predicted chess was finished. For a while Stockfish made top-level chess dull, because players copied engine-perfect play and ground out long endgames. Then neural-net engines arrived, notably Leela Chess Zero, trained through self-reinforcement in the style of Go AIs. They played aggressive, unconventional moves that beat Stockfish and made chess exciting again. Allebest's summary is that computers first made the game boring and then made it far more exciting.
- **Kids are improving faster:** seven-year-olds now give grandmasters real trouble. Allebest's 18-year-old son taught himself video game development from YouTube and uses AI for research on problems, not for the creative work.
- **No shortcut past practice:** with millions of players at every rating, Chess.com's data gives a plain answer about how people get better: repeat the small building blocks. That means puzzles, reviewing your mistakes, and drilling openings. Tools can make practice easier but can't replace it. Allebest compares it to training a neural net, except human brains can't train in parallel.
- **The "×1,000" rule for habits:** take one ordinary day and multiply it by 1,000. Five puzzles a day becomes 5,000 puzzles. The same applies to a daily sketch or a short daily run. Allebest says habits that compound matter more than any single day.
- **Cheating:** Allebest calls it a long-standing problem that is now an arms race, not a threat to the game's existence. Chess.com spends heavily on it, using statistical and machine-learning models that compare how humans play with how engines play. He won't describe the methods. Two separate workstreams handle casual play and prize-money events. He says cheating can't be prevented, since you can't put a webcam in every home, but it can be caught, banned and educated against.
- **Why chess lasts:** Allebest compares it to people returning to the Beatles and Led Zeppelin when there's an endless supply of music. With AI able to produce 10,000 new board games overnight, a classic with no loot boxes, no skins, no rule changes, no luck and full information becomes a shared experience people can rely on. It is simple enough to learn and deep enough to keep surprising.
- **Identity and expansion:** Allebest calls Chess.com a chess company that is also a gaming company, a content company and a social network. Chess is 99.99% of the business. It recently launched **Gambit** (gambit.com, plus mobile), a poker product built around skill ratings rather than money. A small team is working on more classic games.
- **Poker ratings:** the current algorithm weighs the ratings of the other players at the table, chips won, and hands played. Allebest expects to keep adjusting it. The hard part is that behavior changes with the stakes, from $100 games to $1,000 games to playing for M&M's, where someone goes all-in every hand and just rebuys. He thinks players will come to care about their rating as much as money. Losing $100 doesn't bother him much, but losing 100 rating points does.
- **How Chess.com uses AI internally:**
  - Support automation, with human staff building the automations and handling cases that need a person.
  - LLM-assisted analytics, so analysts can ask questions of the data.
  - **GNS**, an internal authentication and knowledge layer across the whole organization with permissions built in, started early by CTO Josh.
  - An in-house martech stack.
  - Agentic development and AI-written specs. The aim is to shrink the time between spotting a problem or opportunity and shipping code by cutting the handoffs between scoping, roadmap, build, QA and release.
- **AI in the product:** aggregated "advanced stats" that compare you with players like you and players just above you, to show what to work on next. An AI coach "in your pocket" for on-demand help, which Allebest says doesn't replace a human coach. AI game review. Personalized puzzles. A prototype coach that reviews your past week of games, gives feedback, and can be chatted with.
- **On AGI/ASI:** Allebest believes superhuman intelligence is coming and speeding up. He leans optimistic, but sees concentrated wealth and poor leadership as the main risks, calling it more a cultural problem than a technical one. He compares it to nuclear technology and guns: outcomes depend on who controls them.

## Actionable Takeaways
1. **Fund growth from revenue when no big competitor is bearing down on you.** In a pure software and content business without hardware or inventory, charge early (Chess.com did it within about 18 months of launch) and add each hire only when revenue covers it. Allebest says this slower route built the company's culture and removed the need for outside capital.
2. **Win a market full of copycats on experience.** If your core can't be patented, compete on what players want: free to play, works in the browser, feels great. Then build community and content around it.
3. **Widen who belongs.** Change the definition of a real participant, as Chess.com did when it treated a 500-rated player as a real chess player, and celebrate beginners' mistakes instead of gatekeeping by skill level.
4. **Don't trust one spike.** Watch whether the baseline after a surge settles higher than before. Chess.com waited through the 2020–22 and 2023 waves before deciding in 2024 that the growth was durable.
5. **If you take PE money, pick partners who back the mission and buy secondary.** Look for operational help with forecasting, reporting and experimentation that improves how you run the company, not just reports for investors.
6. **Get better through daily repetition.** Choose a small daily amount (for example 5 puzzles), multiply it by 1,000 days, and review your own mistakes. Use AI to research and refine ideas, but watch for flattery and bias.
7. **Run anti-cheating as two tracks:** broad statistical and ML detection for casual play, and a stricter process for prize-money events. Keep your detection methods private, and pair bans with education.
8. **Cut handoffs from idea to ship.** Use AI-written specs and agentic development so that spotting a problem leads straight to a fix, collapsing the scope → roadmap → build → QA → release chain.
9. **Use ratings to measure skill where money currently does.** Gambit's model for poker rates players on opponents' ratings, chips won, and hands played. The aim is to separate real skill from deep pockets or bullying.
10. **For founders unsure about their market:** decide what you want to exist, find the smallest set of resources that can prove it, pick great people to work with, listen to customers and pivot if it isn't working, and ignore the flood of generic founder advice.

## Notable Quotes
> "Oh, this might just be a flash in the pan like treadmills and sourdough bread." — Erik Allebest

> "Most of them said this is uninvestable. You should get a real job." — Erik Allebest

> "Computers at first made chess a little bit more boring, but then made it so much more exciting." — Erik Allebest

> "You can't really shortcut practice. You have to put in the time." — Erik Allebest

> "There's no loot boxes, there's no new skins that dropped, there's no like changes to the rules." — Erik Allebest

> "Losing 100 rating points on my poker, that just like really bothers me." — Erik Allebest

> "Stop listening to people just giving you advice on what to do. Just go do it." — Erik Allebest

---

*You're reading a protocol on [protocol.supply](https://protocol.supply) — hosted, searchable knowledge for AI agents. Think Agent Skills, as an API: bootstrap once (`curl -sL https://api.protocol.supply/bootstrap`), search (`/search?q=…`), fetch (`/@repo/slug`).*

*Any skill can become a protocol, and any protocol installs as a local skill — tell your agent "install this as a skill" or see [`@protocol/protocol-to-skill`](https://api.protocol.supply/@protocol/protocol-to-skill).*