A Raspberry Pi blog post that became a US$1 million-a-year hardware company, sold for US$598,000

Michael worked at Google as a software developer for four years and quit on 1 February 2018. 4 Two tries at promotion had failed. The second time, the committee said six months of senior-level work was not a long enough record. 4

By then Michael had found Indie Hackers, a community of founders who build small, profitable businesses without investors. Michael decided to become an indie hacker too, with no business idea yet: "I am indeed an idiot with no idea." 4

The first year lost US$20,871. 5 Zestful, an API that reads recipe ingredients, earned nothing in 2018. Space Duck, built on the Sia storage network, was shelved in April. Is It Keto, a site that says which foods fit the keto diet, had just started earning from affiliate links. 6

The second year lost US$2,402, on US$7,254 of revenue. 5 In 2020 more ideas failed. WanderJest, a guide to live comedy shows, closed when COVID hit. Portfolio Rebalancer drew 1,000 visitors in its first month, and one free-trial sign-up. 7 In two years Michael earned less than US$10,000 in total. 7

Michael ran a home server with no monitor or keyboard. When a mistake locked it out, the fix was to drag it to the desk and swap cables. Commercial devices that solve this cost US$500 to US$1,000. 1

So Michael built one. A Raspberry Pi, a small computer that costs US$30 to US$60, pretends to be a USB keyboard. An HDMI-to-USB dongle bought on eBay for US$11 captures the screen. A web page shows both. 1 With help from Max Devaev, who wrote the uStreamer video tool, the delay fell from about 10 seconds to 200 milliseconds. 1

Michael doubted anyone would pay for a kit of parts anyone can buy: maybe one or two orders a week, at about US$80 a kit. 7 Before the blog post, two kits had sold. 8

The post went up on 23 July 2020. 1 It reached #1 on Hacker News and stayed there most of the day. 8 In less than four hours, all nine kits in stock were sold, and people kept buying while the kit was backordered. 7

July ended with 52 kits sold and US$8,741.37 in revenue, against US$173.94 in June. Margins were around 40-50%. 8 In Michael's words, TinyPilot "was product/market fit at first sight." 7

The reason it worked, by Michael's account: for the first time, the blog's readers overlapped strongly with the product's likely customers. 8 In November 2020 a higher-end model sold 55 units in its first month, about US$14,000 of extra revenue. 7

At the start of 2021, Michael wrote the code and handled support, sales and marketing. One other person managed stock and packed orders, and the house was slowly turning into a warehouse. 9

By early 2022 TinyPilot had an office, two people for shipping, three developers and a distributor in Europe. 9 In October 2021 a member of the shipping staff began sharing customer support, and soon answered about 70% of support emails alone. 9

The same month, Michael removed the product catalog page and sold only the main product, the TinyPilot Voyager. Monthly revenue went from US$20,000-30,000 to US$40,000-60,000, with no new reviews and no marketing. 9

By the end of 2022 the team was seven: Michael, two part-time developers, two part-time staff for assembly and shipping, and two part-time support engineers. Support fell from about 20% of Michael's time to under 5%. 10 The rule: every role works at about 50% capacity, half on urgent work, half on planned improvements. 10

In February 2023 the 3D-printed case became a metal case. Michael raised the price 10%, and monthly sales were still higher than before. 3 In March 2023 shipping moved to an outside warehouse, and in the second half of the year a contract manufacturer took over production. 3

Sales grew every year, but profit did not. In 2022 TinyPilot sold US$807,458 and kept US$10,447. 3

The cause was the global chip shortage. It forced frequent redesigns of the circuit board, and the new electrical engineering firm cost three times as much as the old one. Electrical engineering cost US$124,643 that year. 10

In 2023 Michael stopped changing the hardware. Electrical engineering fell to US$23,427, and the metal case lifted both price and sales. Profit rose to US$235,568, 20 times the year before. 3

Michael drew no salary, so that profit was the full year's pay. 3

The lesson Michael drew that year: "Success is more stressful than failure." 7

In the review of 2021, Michael asked the yearly question, whether the work was still a joy, and for the first time found it hard to say yes. 9 2022 was the hardest year: with the chip shortage, no two batches could be built the same way. 10

In February 2023 a sprinkler burst in the room next to the office, where all the stock was kept. The stock stayed dry, but Michael decided never to depend on one small office again. 3

Then the reasons to sell. Michael missed writing code and managed six people on three teams alone. TinyPilot "occupied about 40% of my waking hours, but it occupied 90% of my stress." 2

In October 2022 Michael asked FE International, a broker, about a sale. The answer was no: two products made about 98% of revenue, and buyers usually want at least 10, none above 30%. 2

So Michael made the business easier to run instead: no more hardware redesigns, and the office's work moved to outside companies. That also made it more profitable, and a new owner could now run it from anywhere. 2

Michael wrote to the CEOs of five companies whose products fit next to TinyPilot. One offered US$150,000 in cash, a one-year job at US$100,000 and a share of profit for two years. The cash was 0.9 times earnings, and Michael said no. 2

At Microconf 2023 Michael met Quiet Light, a broker. TinyPilot was listed at US$599,000, about 2.9 times the last twelve months of earnings, US$208,000. 2 The buyer found TinyPilot there and then read Michael's public retrospectives. 2

The buyer first offered US$500,000, and Michael declined. A second pair of buyers dropped out. The next day the buyer offered the asking price, and Michael accepted. 2

The buyer used a bank loan backed by the US Small Business Administration, so it took three months from the letter of intent to closing. In those months Michael spent half the working time on the sale. 11

Everyone on the team was offered a job by the new owner, and all accepted. 2

In May 2024 a Google recruiter offered Michael the old job back with no interview, and Michael was not tempted. TinyPilot still sells KVM devices today, the Voyager 3 and the Pathfinder. 12

1. How much of TinyPilot's revenue came from Europe through the distributor?
2. Did the custom board Michael planned for 2022 become the Voyager 3 sold today?
3. Looking back, would Michael build a hardware business again, or only software?
4. How did the team's work change in the first year under the new owner?