Super AI Singapore: Through Our Eyes
An unfiltered look at the technologies, and takeaways from Super AI 2025
Asia Tech Lens was on the ground this week at Marina Bay Sands for Super AI Singapore 2025, an event that pulled in over 7,000 attendees from more than 100 countries. While we knew it was going to be a well-attended event, the enormity only hit us when we arrived on site. What struck us most was the diversity of who showed up. You had the big players, cloud giants, platform providers, robotics companies, side by side with startups trying to solve very specific problems. It was a reminder that AI isn't just about moonshots, it's about the quiet, practical stuff too.
Robots Steal the Spotlight
As we walked into the event, the first reaction was definitely “woah”! It almost felt like we had entered a star wars movie set. The atmosphere was electrifying, and there were robots! A bunch of them were playing chess, one was sketching portraits, and others were weaving through the crowd like they belonged there.
It was playful, yes, but it was also a glimpse of how quickly physical AI is moving into real-world environments.
So it wasn’t surprising that the panel on “Physical AI: The New Age of Intelligent Robotics” drew such attention. Moderated by Auki Labs’ Nils Pihl, it brought together Felix Shang from Unitree and Nicolaus Radford from Persona AI for a lively, and at times candid, conversation on where humanoid robots are really headed.
The panelists made one thing clear, the industry is still in infancy. But the big question was - where will these robots be scaled first?
Some said homes but that’s, at least, five years away. Homes are unpredictable - every layout is different, every mess is unique. Not exactly ideal conditions for a robot still learning how to climb stairs. Others argued that the smart money is on heavy industry. Think factories, warehouses, welding floors, places where tasks are repetitive or dangerous, the ROI is obvious, and the labor shortages are common.
A lot of you might think that the U.S. would be leading the way. But then you would be wrong.
It’s Asia that’s in the lead. The region accounted for 73% of global industrial robot installations in 2022, led by China, Japan, and South Korea. China installed more robots last year than the rest of the world combined. South Korea has one robot for every 10 manufacturing workers. Singapore, despite its size, ranks number one globally for robot density. it has more robots per 10,000 employees in the manufacturing industry than any other country, according to the IFR International Federation of Robotics
By comparison the U.S. doesn’t even crack the top 10. Part of the reason is structural: the regulatory environment is more rigid, and the relationship between labor and manufacturing is harder to untangle. The mood wasn't defeatist, though. The view from the ground was that the U.S. is not out, it just needs to solve the problem differently.
AI Eats the World
In his session, AI analyst Benedict Evans framed the bigger picture: we're still figuring out whether generative AI is merely the next platform shift (like smartphones after PCs) or something more fundamental, like electricity.
The irony is stark: tech giants will be investing $300B in infrastructure this year alone (Microsoft now spends 30% of revenue on capex), yet surveys show only 10% of people use tools like ChatGPT daily.
Evans reframed this disconnect with a telling analogy: "We’re in the VisiCalc moment. Accountants in 1978 saw spreadsheets revolutionize their work overnight. Everyone else just saw a blank grid."
Surveillance in the Age of AI: Snowden’s Warning
Among the most crowded sessions we attended at Super AI was one featuring Edward Snowden - probably one of the best people to talk about privacy and freedom in the age of AI. Joining remotely, the former NSA contractor-turned-whistleblower delivered a cautionary address about the growing power of AI in areas of surveillance, and evaluation.
Snowden began by charting how far machine perception has come since 2013. Back then, analyzing a minute of video could take an hour of processing. Today, AI models can process 30 hours of footage in just one hour. Surveillance, once constrained by human limitations, is becoming real-time - capable of recognizing faces, voices, and movement patterns.
But Snowden’s warning wasn’t just about privacy, it was also about autonomy. He painted a stark picture of AI systems that don’t just observe our behavior, but gradually start nudging, flagging, and even steering it. More often, not for the benefit of the individual, but to push toward an “average”, defined by whoever controls the system.
He warned that progress rarely emerges from the norm, it is born from deviation, from people who challenge the status quo.
“The average is not the ideal,” he said. “The average is the enemy. It’s unoriginal.”
“We need technology to help the community escape away from the average and towards our finest moments. What I fear is that an ignorance of this will more likely, instead, to chain us toward the weight of our past mistakes.”
For More Info on Asia Tech Lens