Factory Floor Software: How Ex-SpaceX Engineers Are Changing Manufacturing
Two former SpaceX engineers have taken the software that helped put rockets into orbit and reimagined it for a completely different mission — the factory floor. Their startup, Sift, is quietly becoming one of the most important players in the movement to make physical manufacturing smarter, faster, and ready for the age of artificial intelligence. If you have ever wondered what it actually takes to automate a factory, the answer starts with data.
 |
| Credit: Sift Stack |
The "Atoms, Not Bits" Era Has a Data Problem Nobody Is Talking About
Silicon Valley has spent the last few years falling back in love with physical things. Rockets, chips, electric vehicles, satellites — there is a renewed obsession with building hardware, not just software. The rallying cry of "atoms, not bits" has grown so loud that major investors are reportedly assembling enormous funds to buy and modernize old manufacturing firms using artificial intelligence.
But here is what most people miss when they talk about automating factories: the hardware transformation is completely impossible without a software revolution happening underneath it. Machines do not run themselves. They generate an almost incomprehensible volume of data, and someone — or something — has to make sense of all of it in real time. That is exactly where Sift has staked its claim.
From Rocket Launches to Factory Lines: The Origin Story of Sift
Karthik Gollapudi and Austin Spiegel both worked at SpaceX before founding Sift in 2022. Their jobs there revolved around a very specific challenge: managing the torrents of telemetry data that rocket components stream during testing, manufacturing, and launch. Telemetry data is the real-time performance information that sensors on physical machines constantly broadcast — temperature, pressure, vibration, velocity, and thousands of other signals all firing at once.
When they looked at how other companies building complex machines handled this kind of data, what they found was underwhelming. Most organizations were either patching together off-the-shelf database tools not designed for this purpose, or writing custom Python scripts internally. Neither approach scaled well. Neither approach was truly ready for what AI would eventually demand from manufacturing data. Sift was built to close that gap.
Why Sift's Customers Are Dealing With Data Volumes That Would Shock You
The numbers involved in modern industrial data are difficult to wrap your head around. Some of the vehicles that Sift works with have more than 1.5 million sensors streaming information simultaneously, across multiple formats and time scales. That is not a typo. Over a million sensors, all generating data at the same time, all needing to be captured, organized, and stored in a way that is actually useful.
One satellite company that relies on Sift for test management, manufacturing oversight, and operations can run as many as ten million automated software tests in a single day. The cost of storing all that data without a smart infrastructure strategy quickly runs into millions of dollars per month. The question is not just whether you can store it — it is whether you are spending that storage budget on data that is actually going to make your machines better. That distinction is everything.
AI Changed the Rules — And Sift Adapted Faster Than Anyone Expected
Gollapudi is candid about the fact that the arrival of modern AI tools forced Sift to rethink its strategy. When the company launched, its signature strength was building customized data workflows — bespoke pipelines tailored to each client's specific needs. That was valuable in 2022. By 2025, it had become table stakes. AI and deep learning models had made powerful data analysis accessible enough that customized workflows alone were no longer a meaningful differentiator.
What had become more valuable — dramatically more valuable — was the underlying data infrastructure itself. The ability to collect data from wildly different sensors and systems, normalize it, store it efficiently, and then make it readable by AI agents and machine learning models: that is the hard problem that most manufacturers have not solved. Gollapudi describes Sift's mission as making that data "machine readable." If AI is going to flag a production defect before it becomes a recall, or optimize a test sequence to cut costs, the data has to be clean, organized, and accessible. Sift's entire business is built on that premise.
The Customers Already Trusting Sift With Their Most Critical Systems
Sift's client roster reveals just how serious the company's ambitions are. On the aerospace side, it works with major rocket builders and defense contractors who cannot afford data failures. On the startup side, it serves robotics companies and power grid management platforms — industries where real-time sensor data is not a nice-to-have, it is operationally critical.
The satellite industry is a particularly revealing case. When a company runs ten million automated tests a day, data infrastructure is not a back-office concern. It is a frontline competitive advantage. Companies in this space have told Sift directly that smart data management has taken the worry out of storage costs entirely — replacing anxiety about runaway cloud bills with confidence that every dollar spent on data is generating real value.
Sift Just Raised $42 Million — Here Is What That Signals for Industrial AI
In 2025, Sift closed a $42 million Series B funding round at a post-money valuation of $274 million. The round was led by StepStone, with participation from GV — the venture arm of Google — along with Riot Ventures, Fika Ventures, and CIV. The involvement of Google's venture arm is particularly telling. It signals that the largest technology companies in the world see industrial data infrastructure as a critical battleground, not a niche market.
The funding positions Sift to expand both its platform capabilities and its customer base at a moment when the demand for factory-floor AI is accelerating rapidly. Manufacturing companies that once treated data as a byproduct of production are beginning to treat it as a strategic asset. Sift is betting that it can be the company that helps them make that transition.
What the Factory of the Future Actually Looks Like
The factory of the future is not just a building full of robots. It is a living, breathing data ecosystem where every machine, every sensor, and every production run generates information that feeds into AI systems capable of making real-time decisions. Predictive maintenance. Automated quality control. Dynamic supply chain adjustments. None of that works without a rock-solid data foundation underneath it.
Sift's founders understood something early that much of the industry is only now catching up to: the companies that win the manufacturing AI race will not necessarily be the ones with the best robots. They will be the ones with the best data. That insight, forged during years of working on software for some of the most demanding engineering challenges on the planet, is what makes Sift's story worth watching closely.
The ground is shifting fast beneath every company that builds physical things. Sift is betting it knows exactly what the new foundation needs to look like — and the investors backing it seem to agree.
