The rallying cry of “atoms, not bits!” — a phrase that reflects Silicon Valley’s increasing focus on physical manufacturing in contrast to digital products — escalated last week with news that Jeff Bezos is assembling a $100 billion fund to consolidate and automate factories.

However, automating factories is not solely a hardware issue. It is becoming increasingly reliant on advanced software and AI applications, and this transition is transforming the firms developing the infrastructure of the physical manufacturing sector.

Karthik Gollapudi, the CEO of Sift, a company based in El Segundo, California, whose tools facilitate the design and manufacturing of intricate machinery like spacecraft and automobiles, is sensing substantial changes on the horizon. He noted that these developments have altered his company’s direction in the past six months.

Gollapudi and his co-founder, CTO Austin Spiegel, launched the company in 2022 after their experience developing software tools at SpaceX that handled the enormous volume of telemetry data — real-time performance metrics streamed from sensors on physical components — during testing, production, and launch.

While most companies creating advanced machines utilize standard database tools or develop their own Python scripts, Sift identified the chance to offer businesses a top-tier solution. Their clients include United Launch Alliance, a prominent US rocket manufacturer, as well as other defense contractors, robotics, and power grid management startups.

Nonetheless, Gollapudi indicates that the onset of AI tools for data examination prompted a shift in his company. The tailored workflows that used to differentiate the company’s signature offerings have become essential in a landscape dominated by AI and deep learning models. Yet, the company’s talent in managing data infrastructure has unexpectedly gained significance.

“Our long-term vision of how we expected this to unfold over five years is actually materializing this year,” Gollapudi informed TechCrunch. 

This entails managing the significant data flow from modern software-heavy machines. Certain vehicles that the company collaborates with feature over 1.5 million sensors transmitting data simultaneously, in various formats and time intervals. 

The objective of organizing and archiving that data for AI usage is key for the company—”much of the value lies in making that machine-readable,” Gollapudi stated. For AI agents to make manufacturing decisions or analyze test data to identify potential issues, Sift aims to ensure that the data is accessible to them.

Jeff Dexter, the VP of software at Astranis, a satellite company utilizing Sift to oversee testing, manufacturing, and operations, remarked that robust data infrastructure is crucial for businesses like his that might conduct 10 million automated software tests daily. 

“Eventually, it becomes a situation where it’s costing us millions of dollars every month just to retain data,” Dexter expressed. “It’s essentially a question of whether this million dollars is well spent? With solutions like Sift, I don’t have to be concerned about the volume of data present.”

Gollapudi mentioned to TechCrunch that Sift secured a $42 million Series B in 2025 at a post-money valuation of $274 million, led by StepStone with contributions from GV (Google’s venture division), Riot Ventures, Fika Ventures, and CIV.