With Sift Stack, two ex-SpaceX engineers are bringing the software that helped launch rockets to the factory floor

The outcry of “atoms, not bits!” — a operation capturing Silicon Valley’s increasing obsession with carnal manufacturing implicit integer products — reached a fever transportation past week with connection that Jeff Bezos is putting unneurotic a $100 billion money to rotation up and automate factories.

But automating factories isn’t purely a hardware problem. It progressively depends connected blase bundle and AI tools, and that displacement is reshaping the companies gathering the infrastructure of the carnal manufacturing world.

Karthik Gollapudi, the CEO of Sift Stack, an El Segundo, California, institution whose tools enactment the plan and manufacturing of analyzable machines similar spacecraft and cars, is feeling the crushed displacement underfoot. He says these changes person reshaped his company’s absorption successful the past six months.

Gollapudi and his co-founder, CTO Austin Spiegel, started the institution successful 2022 aft moving connected bundle tools astatine SpaceX that managed the immense magnitude of telemetry information — real-time show accusation streamed from sensors connected carnal components — during testing, manufacturing, and launch.

Most companies gathering precocious machines usage off-the-shelf database tools oregon navigator up their ain Python scripts, but Sift saw the accidental to supply companies with a best-in-class tool. Customers scope from United Launch Alliance, a large US rocket builder, and different defence contractors, to robotics and powerfulness grid absorption startups.

However, Gollapudi says that the accomplishment of AI tools for information investigation forced a alteration astatine his business. The kinds of customized workflows that erstwhile stood retired arsenic the company’s signature offering person go array stakes successful a satellite of AI and heavy learning models. But the company’s quality to negociate information infrastructure had abruptly go much valuable.

“Our semipermanent imaginativeness of however we saw this playing retired implicit 5 years is really being played retired this year,” Gollapudi told TechCrunch. 

That means managing the aggravated information travel from today’s software-intensive machines. Some vehicles the institution works with person much than 1.5 cardinal sensors streaming information concurrently, crossed aggregate formats and clip scales. 

Organizing and storing that information for AI applications is the company’s goal—”a batch of the worth is successful exposing that to beryllium instrumentality readable,” Gollapudi said. If AI agents are going to marque decisions astir manufacturing oregon analyse trial information to emblem imaginable problems, Sift’s extremity is to marque that information disposable to them.

Jeff Dexter, the VP of bundle astatine Astranis, a outer institution that uses Sift to negociate test, manufacturing, and operations, said that bully information infrastructure matters for companies similar his that mightiness bash 10 cardinal automated bundle tests successful a day. 

“Inevitably, it gets to a constituent wherever it’s costing america millions of dollars per period conscionable to store data,” Dexter said. “It’s truly like, is this a cardinal dollars good spent? With exertion similar Sift, I don’t interest astir however overmuch information is there.”

Gollapudi told TechCrunch that Sift raised a $42 cardinal Series B successful 2025 astatine $274 cardinal post-money valuation, led by StepStone with information from GV (Google’s task arm), Riot Ventures, Fika Ventures, and CIV.