Bitfusion

Three engineers—Subu Ramat, one from Samsung, and one from Intel—were working at some of the world's largest tech companies when they spotted an opportunity in late 2014. They'd each spent time building high-performance systems and cloud infrastructure, and noticed a huge gap: only big enterprises could afford hyper-convergence and supercomputing power. Meanwhile, data centers everywhere were sitting on idle resources. "We saw this trend where only big companies can actually take advantage of high-performance computing," Subu recalls. "We saw the opportunity where how do we bring that to the normals?" The three founders had known each other for roughly eight years, so the decision came fast. They met for lunch at Korea House (pitching to the Samsung guy), then grabbed Mozart's coffee (pitching to the Intel engineer). "In about like three, four months, we were out," Subu says.

Bitfusion's core insight was elegant: just as you could combine processing power from multiple smartphones to make one faster, they could pool unused compute resources across enterprise data centers to create single, incredibly powerful instances. They called it software-defined compute—applying the same philosophy that had worked for networking and storage. The three founders collectively gave up salaries around $250,000 each to bet on the idea. "We basically make a bunch of commodity servers, make it look like a supercomputer," Subu explains. They spent nine months engineering the technology, filing eight patents along the way, and building what they knew would be a hard-to-replicate product. By the time they launched their product in November 2015, they had a deep technical foundation ready for the market.

Rather than chase the traditional enterprise sales cycle (6-9 months), Bitfusion did something clever: they worked backwards. They created customer personas for companies that could close in 1-3 months, then targeted those segments aggressively. The strategy worked faster than expected. "We basically like literally in a couple of months, we basically have six guys, you know, essentially we actually are doing pilots with six guys and we have two contracts signed," Subu says. In just three months post-launch, they'd landed customers ranging from cloud service providers all the way up to major oil companies like Shell. The pricing model was flexible—some contracts were $10,000/year, others hit $100,000/year.

The team's engineering-first approach worked. Nine people, nearly all engineers, meant they could ship fast and iterate on product. By April 2015—before they even had meaningful revenue—they raised $1.5M in seed funding through Y Combinator's SAFE framework in just two weeks. They went live with their funding round the first week of April and closed it by mid-April. "Fundraising was the most easiest thing. I never thought it would be that easy," Subu says. Their seed investors included Data Collective, a Silicon Valley VC, plus investors in Texas, all of whom brought strategic value beyond just capital. What didn't work at first was the traditional enterprise sales motion—so they ditched it and created their own playbook. Their burn rate stayed lean: less than $100,000 per month even while paying market-rate engineers.

By the time of this interview (mid-2016, roughly 18 months after launch), Bitfusion had six people in Austin and a distributed team in Europe. They'd generated somewhere between $100k and $500k in total revenue—Subu wouldn't disclose the exact figure. The ambition was clear: by end of 2016, they wanted enough enterprise customers to fund a Series A, even though investors were already circling. Subu wasn't chasing a particular valuation; he was hunting for the right strategic partner. "It's all about the right strategic investor," he says. His north star was VMware, which had turned software-defined storage and networking into a $100B+ company. "What people did in software defined networking and software defined storage, we are trying to do the same for software defined compute. We think this could be as big as VMware, if we executed right in the next few years."