The cry of “atoms, not bits!” — a phrase capturing Silicon Valley’s rising obsession with bodily manufacturing over digital merchandise — reached a fever pitch final week with phrase that Jeff Bezos is placing collectively a $100 billion fund to roll up and automate factories.
However automating factories isn’t purely a {hardware} downside. It more and more will depend on subtle software program and AI instruments, and that shift is reshaping the businesses constructing the infrastructure of the bodily manufacturing world.
Karthik Gollapudi, the CEO of Sift Stack, an El Segundo, California, firm whose instruments assist the design and manufacturing of complicated machines like spacecraft and vehicles, is feeling the bottom shift underfoot. He says these adjustments have reshaped his firm’s focus within the final six months.
Gollapudi and his co-founder, CTO Austin Spiegel, began the corporate in 2022 after engaged on software program instruments at SpaceX that managed the large quantity of telemetry knowledge — real-time efficiency info streamed from sensors on bodily parts — throughout testing, manufacturing, and launch.
Most firms constructing superior machines use off-the-shelf database instruments or prepare dinner up their very own Python scripts, however Sift noticed the chance to supply firms with a best-in-class software. Prospects vary from United Launch Alliance, a significant US rocket builder, and different protection contractors, to robotics and energy grid administration startups.
Nonetheless, Gollapudi says that the arrival of AI instruments for knowledge evaluation pressured a change at his enterprise. The sorts of custom-made workflows that when stood out as the corporate’s signature providing have grow to be desk stakes in a world of AI and deep studying fashions. However the firm’s capability to handle knowledge infrastructure had out of the blue grow to be extra precious.
“Our long-term imaginative and prescient of how we noticed this taking part in out over 5 years is definitely being performed out this yr,” Gollapudi instructed TechCrunch.
Techcrunch occasion
San Francisco, CA
|
October 13-15, 2026
Which means managing the extreme knowledge circulation from right now’s software-intensive machines. Some automobiles the corporate works with have greater than 1.5 million sensors streaming knowledge concurrently, throughout a number of codecs and time scales.
Organizing and storing that knowledge for AI functions is the corporate’s aim—”quite a lot of the worth is in exposing that to be machine readable,” Gollapudi mentioned. If AI brokers are going to make selections about manufacturing or analyze check knowledge to flag potential issues, Sift’s aim is to make that knowledge out there to them.
Jeff Dexter, the VP of software program at Astranis, a satellite tv for pc firm that makes use of Sift to handle check, manufacturing, and operations, mentioned that good knowledge infrastructure issues for firms like his that may do 10 million automated software program assessments in a day.
“Inevitably, it will get to some extent the place it’s costing us hundreds of thousands of {dollars} per thirty days simply to retailer knowledge,” Dexter mentioned. “It’s actually like, is that this 1,000,000 {dollars} properly spent? With know-how like Sift, I don’t fear about how a lot knowledge is there.”
Gollapudi instructed TechCrunch that Sift raised a $42 million Collection B in 2025 at $274 million post-money valuation, led by StepStone with participation from GV (Google’s enterprise arm), Riot Ventures, Fika Ventures, and CIV.
