When software-intensive machines get it wrong, the price can be catastrophic. Boeing's Starliner spacecraft encountered a software error that sent it into the wrong orbit, costing the company $600 million. That failure stuck with Karthik Gollapudi and Austin Spiegel, who had both worked on software tools at SpaceX that managed telemetry data—real-time performance information streamed from sensors on physical components.
Rather than brush it off as bad luck, they recognised a pattern. From the ispace lunar lander crash in 2023 to the East Palestine train derailment, failures of complex machines have become routine, stemming not from a lack of engineering talent but from an inability to see and understand machine behaviour at scale. These weren't exotic hardware problems. They were data problems.
Gollapudi and Spiegel started Sift Stack in 2022 after seeing the gap between what engineers needed and what existed in the market. Most companies building advanced machines rely on off-the-shelf database tools or write their own Python scripts, but Sift saw the opportunity to provide a purpose-built solution. Today, the company serves customers across aerospace, defence, and transportation.
The core problem sounds mundane until you think about scale. Some vehicles the company works with have more than 1.5 million sensors streaming data concurrently across multiple formats and time scales. Trying to manually review that volume of information is impossible. A satellite company customer runs 10 million automated software tests in a day; storing that data costs millions of dollars monthly, but with Sift, engineers stop worrying about data volume.
Organising and storing data for AI applications is Sift's primary goal, making data machine-readable so AI agents can make decisions about manufacturing or analyse test data to flag potential problems. This matters because machines are becoming increasingly complex, with satellites, hypersonic vehicles, and autonomous robotics functioning as software-driven, data-intensive ecosystems.
The company has traction with serious customers. They range from United Launch Alliance, a major US rocket builder, to robotics and power grid management startups. Sift raised a $42 million Series B in 2025 at a $274 million post-money valuation, led by StepStone with participation from Google's venture arm, Riot Ventures, Fika Ventures, and CIV.
But here's what's worth watching: the scale of the opportunity has expanded far beyond Spiegel and Gollapudi's original idea. Customised workflows that once set Sift apart have become standard in AI and machine learning; instead, the company's ability to manage data infrastructure became suddenly more valuable, with their five-year vision playing out in a single year. That's either tremendous validation of the market timing, or a sign that the category itself is shifting faster than anyone expected.