Real demand from power users. The first users were people who found Chrome slow: Figma, Google Drive, hundreds of tabs and old laptops (Chase Lambert). People would pay for speed. The passionate users were "mainly compelled by its speed" (Chase Lambert). A clean stop. "If it's not going to work, I won't waste our time or money," Doshi wrote, which left half the funding for the next bet (Hackernoon thread archive).

Mighty: why a $35/month cloud browser shut down after 3.5 years

Q: What didn't work for Mighty, and what are the risks?

Not enough gain. "The web is bound by single core performance of JS… We could improve things by 2x but not 5-10x. Ok business, not mass market changing" (Hackernoon thread archive). Churn. Users left over "networking stutters, stability issues, and missing features (like webcam support)", and in some cases it ran slower than Chrome (Chase Lambert). Price expectations. People are used to browsers being free (Liliputing). The original thread was on X; the copy linked here is a Hackernoon archive of it. Revenue and user counts were never published.

Key takeaways

Nov 2022: Suhail Doshi stopped Mighty after 3.5 years: 'We could improve things by 2x but not 5-10x.'
Users paid about $35/month but churned over network stutters, stability issues and missing features.
About 50% of the money raised was left; the team pivoted to AI creative tools (Playground AI).

What it is

Mighty was a Chrome browser that ran on a powerful server in the cloud and streamed to your computer, with the goal of making web apps feel faster (Liliputing). An engineer on the team says people paid "$35/month when the competing products were free" because they valued their time (Chase Lambert).

Timeline

~2019: work begins; Doshi later says he worked on it for 3.5 years (Hackernoon thread archive).
2021: early access program announced, about a year and a half before the shutdown (Liliputing).
13 Nov 2022: Doshi announces he has stopped working on Mighty and offers the IP for sale (Hackernoon thread archive).
Same thread: "we have about 50% of all money raised left," and the team pivots to AI creative tools, introducing Playground AI (Hackernoon thread archive).
15 Nov 2022: engineer Chase Lambert publishes a retrospective (Chase Lambert).

Why it worked

It didn't, commercially. What went right:

Real demand from power users. The first users were people who found Chrome slow: Figma, Google Drive, hundreds of tabs and old laptops (Chase Lambert).
People would pay for speed. The passionate users were "mainly compelled by its speed" (Chase Lambert).
A clean stop. "If it's not going to work, I won't waste our time or money," Doshi wrote, which left half the funding for the next bet (Hackernoon thread archive).

What to learn

Check which way the hardware curve is moving. Apple's M1 had single-core performance "faster than almost any server CPU money can buy" (Chase Lambert).
Measure whether a headline feature is used. When users got 32GB of RAM, "90+% of them didn't go beyond 16" (Chase Lambert).
Rebuilding a mature product (a browser) means rebuilding every ordinary feature: audio, downloads, scrolling, VPN, YubiKey and more (Chase Lambert).
Decide to stop while you still have money to try something else.

Risks and what didn't work

Not enough gain. "The web is bound by single core performance of JS… We could improve things by 2x but not 5-10x. Ok business, not mass market changing" (Hackernoon thread archive).
Churn. Users left over "networking stutters, stability issues, and missing features (like webcam support)", and in some cases it ran slower than Chrome (Chase Lambert).
Price expectations. People are used to browsers being free (Liliputing).
The original thread was on X; the copy linked here is a Hackernoon archive of it. Revenue and user counts were never published.

FAQ

Short answers drawn from the sections above.

What is Mighty?

Why did Mighty work?

Real demand from power users. The first users were people who found Chrome slow: Figma, Google Drive, hundreds of tabs and old laptops (Chase Lambert).
People would pay for speed. The passionate users were "mainly compelled by its speed" (Chase Lambert).
A clean stop. "If it's not going to work, I won't waste our time or money," Doshi wrote, which left half the funding for the next bet (Hackernoon thread archive).

What can indie builders learn from Mighty?

Check which way the hardware curve is moving. Apple's M1 had single-core performance "faster than almost any server CPU money can buy" (Chase Lambert).
Measure whether a headline feature is used. When users got 32GB of RAM, "90+% of them didn't go beyond 16" (Chase Lambert).
Rebuilding a mature product (a browser) means rebuilding every ordinary feature: audio, downloads, scrolling, VPN, YubiKey and more (Chase Lambert).
Decide to stop while you still have money to try something else.

What didn't work for Mighty, and what are the risks?

Not enough gain. "The web is bound by single core performance of JS… We could improve things by 2x but not 5-10x. Ok business, not mass market changing" (Hackernoon thread archive).
Churn. Users left over "networking stutters, stability issues, and missing features (like webcam support)", and in some cases it ran slower than Chrome (Chase Lambert).
Price expectations. People are used to browsers being free (Liliputing).
The original thread was on X; the copy linked here is a Hackernoon archive of it. Revenue and user counts were never published.

Sources

Every link cited above, in order of first use.

1.Liliputing (liliputing.com)
2.Chase Lambert (blog.chaselambda.com)
3.Hackernoon thread archive (hackernoon.com)