Last week, the founding team of SAVA Robotics, a manufacturing robotics startup out of Y Combinator’s 2025 spring batch, officially unveiled their first product, the CNC Press Brake, to the public. SAVA doesn’t just have the potential to become a unicorn company. The startup, with plans to launch fully-autonomous factories, could entirely transform the United States’ production capabilities and close the gap between East Asian competition.
“SAVA is here to bring the supply of AI-driven and data-informed robotics to manufacturing. American manufacturing can’t simply throw lots of money on inflexible, clunky hardware that replaces existing brick and mortar shops,” co-founder Alessio Toniolo told the Perrin Institution. “We need to build solutions that retrofit existing manufacturing into 50,000 small fabrications.”
How It Started
After SAVA CEO and Georgia Institute of Technology drop-out Jakob Knudsen effectively pitched the company to Y Combinator General Partner Aaron Epstein at a dinner, he and his high school-aged co-founders, Alessio Toniolo and Vedic Patel—who met their junior year of high school at Georgia Tech’s Research Insitute—were ushered into a YC interview and ultimately accepted into the highly-prestigious accelerator. A month later, they released CNC Press Brake, a 250-pound robot model they built in a shared San Francisco apartment.
“Y Combinator started with a rollcoaster, with experts telling us SAVA has no market nor technical viability,” said Toniolo. “Today, SAVA Robotics is focused on bringing manufacturing back onshore by creating technology—designed to integrate with existing infrastructure—that engages human workers rather than replacing them.”
As high school and college drop-outs without long careers of experience, the three SAVA co-founders have explored countless factories throughout San Francisco to better understand the market.
“We have been to every sheet metal shop in the Bay Area, talked to startups in cutting-edge automation, and learned from traditional companies that produce machinery for sheet metal,” said Toniolo.
“We saw automation cells collecting dust and more than half of the machinery not being utilized,” said Knudsen.
The team bore witness to antiquated tools taking up space in factories that lack modern efficiency. Additionally, they came to understand that the American manual labor force is only growing older, and technology needs to fill the gaps left by aging workers.
“Over half of skilled labor in the U.S. is retiring and over the age of 50. As manufacturing is reshored to American soil, we require a new era of versatile, intelligent robots to augment our labor force,” said Toniolo.
In-house automation is not an easy fix for these shops. The time and financial cost of renovating factories and implementing more advanced technology holds many managers back.
“Small shops can’t afford automation: robots need time-consuming programming, causing owners to turn down business,” said Toniolo.
The History of US-China Manufacturing Competition
“The consistent topic that is brought up when we visit shops is China. China is catching up to the U.S. in AI talent (DeepSeek) while still retaining the upper hand in infrastructure,” said Toniolo.
In the post-WWII era and into the 1970s, the United States dominated all tiers of manufacturing, with complete industrial ecosystems and a strong middle-tier manufacturing base. Since the 1980s, the U.S. has shifted its focus to design and software, using foreign labor and resources for raw materials and initial manufacturing while completing the late-stage production process on home soil.
“The outsourcing done in the 80s had its benefits at the time, but now we are paying the price,” said Knudsen.
In this time, China has taken over as a global supplier of manufactured goods, controlling all areas of production as did the U.S. after WWII. Using innovative tools and techniques developed by American companies, China has since become a global leader in product exports. They have, by a wide margin, the highest manufacturing output in the world: over 30% of global output, nearly twice the amount produced by the U.S.
Now, in addition to making advancements in software, China is experimenting with fully automated shops, known as “dark factories,” that function without human labor or supervision. Recognizing the gap between the U.S. and China, the SAVA team understands the advanced utilization of robotics and AI needed to achieve this level of automation, and China’s constant innovation is informing SAVA of where to build next.
SAVA’s First Robot: The CNC Press Brake
SAVA’s first robot, the CNC Press Brake, replaces a technician who would traditionally operate factory machines that form metal sheets.
“The robot uses VLMs—think LLMs with a camera—to process information from the machinery itself. Then it uses a bunch more complex models to recognize and manipulate the parts and this all feeds into the larger system of the factory,” said Patel.
The CNC Press Brake is the first of SAVA’s line of automated robot operators. Next, the team is looking to create robots that can complete other factory tasks, such as welding. All of SAVA’s groundbreaking products will be autonomous workers that make sense of the tasks they are completing, rather than relying on direct commands from human factory workers—saving shop owners significant time for other operational tasks.
“SAVA uses state-of-the-art AI vision models. Our robots see and understand the real world with aerospace-grade precision,” said Toniolo.
The SAVA Future: Fully-Automated Factories
“We are slightly shifting our focus from just the robot that does the one specific process of sheet metal to creating an end-to-end factory for sheet metal that is fully automated and can rapidly scale through acquiring smaller shops,” said Patel. “Now the vision for our company is to basically create Tesla Gigafactories that are fully autonomous for sheet metal through and through.”
Leveraging the edge they have with their advanced robotics technology, SAVA aims to scale through vertical integration. They plan to have autonomous robots that communicate with one another to manufacture parts independently of humans.
“The factory system will give the robot a set of parts to complete, and the robot can use state-of-the-art pose estimation models to recognize and manipulate parts just based off of the CAD. Then it uses VLMs to process the CNC press brake itself to create the part,” said Patel.
“We want to build smarter factories with flexible robots, so humans don’t have to do repetitive manufacturing tasks and can focus on product design, hard engineering problems, and the advanced technology domains that will give America the edge,” said Toniolo.
Although the landscape of robotics technology and artificial intelligence is constantly changing, one thing is certain at SAVA: the team isn’t looking for a quick exit.
“For the team and I, we see this as our life’s work. When you extend the time horizon that far, the perceived risk melts along with it,” said Knudsen.
Watch SAVA’s launch video: https://youtu.be/pnrhTYbOqNc?si=fD0hjffDtvhUZjSX