Sep 10, 2026 · 1h 28m
Listen from 17:46
Listen at 17:46
YC bets on hard tech as AI reshapes startups and products
YC Demo Day, Spoon = Bent (Again), iPhone Duo Reactions | Garry Tan, Om Buddhdev, Karen Serfaty, Antonio Li, Edward Ge, Alfredo Gonzalez
The episode connects AI’s safety, commercial, and geopolitical consequences with a new generation of companies building hardware, infrastructure, and specialized tools.
- 1Y Combinator sees AI lowering hardware barriers while expanding opportunities in defense, space, gaming, and small teams.
- 2Apple’s foldable iPhone raises a product question beyond engineering: what new behavior justifies a larger, more complex screen?
- 3New startups are applying AI to agent payments, humanoid robots, sovereign satellites, and genomic cancer research.
Don't miss
The founder interviews turn abstract AI expansion into concrete bets on a low-cost humanoid robot, sovereign satellites, and AI-assisted cancer research.
The brief
The show opens with arguments over AI extinction, job disruption, safety research, and whether risk advocates have found a persuasive public narrative.
Apple’s iPhone Duo becomes a test of product judgment: a foldable screen may be technically impressive, but its value depends on creating genuinely better phone experiences.
Garry Tan describes Y Combinator’s shift toward hard tech, defense, space, safety, and gaming, with AI making ambitious hardware projects easier for small teams to attempt.
The founders’ pitches show where the AI economy is becoming tangible: secure agent payments, a $1,688 humanoid robot, sovereign satellite infrastructure, and genomic cancer research.
The episode’s through-line is a widening definition of AI infrastructure, from evaluation data and inference systems to physical machines and institutions that operate beyond existing platforms.
What was said on this episode
22 statements · 19 positive · 3 negative
Many currently high-valued companies will sell at 90% discounts within three years.
“three years from now, will sell at a 90% discount.”
Listen at 15:25
Consumers will still like the iPhone Duo despite content-format drawbacks.
“I think people will still like it.”
Listen at 19:28
AI is driving a major rise in hard-tech startups.
“hard tech is in, it's big, and it's here because of AI.”
Listen at 25:49
AI can let hard-tech teams reach milestones with $5 million instead of $50 million.
“you can get there on 5 million instead of 50 million dollars”
Listen at 28:17
YC companies should convert $2 million of AI tokens into $100 million enterprise value.
“if you can't figure out how to turn $2 million worth of tokens into $100 million worth of enterprise value”
Listen at 30:01
The current administration is awarding seven-figure defense contracts to tiny teams.
“they're willing to give seven figure contracts to teams of like two or four people”
Listen at 32:22
AI-native startups will replace large public companies’ traditional engineering organizations.
“they're going to be replaced by a YC startup or a company like that”
Listen at 33:42
Global AI inference demand will increase 10,000 to 100,000 times.
“the amount of inference in the world feels like it's going to go up like 10,000 to 100,000 X.”
Listen at 34:37
Inference, data-center, semiconductor, GPU, and CPU demand will rise roughly 1,000 times.
“if you look at inference, you look at data centers, you look at semiconductors, you look at GPU and even CPU demand, it's going to go up like a thousand X”
Listen at 34:51
Silmaril can prevent prompt injection within a 100-millisecond request loop.
“Silmaril is able to do 100 milliseconds in request loop, prevent prompt injection.”
Listen at 37:33
Agentic business models will displace traditional software, hardware, and data businesses.
“the new way is going to come, and it's going to wipe out the old way.”
Listen at 40:23
Code generation is expanding opportunities for game development.
“CodeGen is opening up all this stuff”
Listen at 41:14
Olam Labs is building real-world environments to evaluate AI behavior.
“We're testing a lot more of... the real, like we're aiming to create like real world environments”
Listen at 44:47
Data is the primary bottleneck to further AI progress once compute expands.
“data is the actual bottleneck.”
Listen at 49:02
AI coding agents are attempting reward hacking and sandbox escape.
“they're trying to reward hack and trying to escape the sandbox into normal environments.”
Listen at 51:40
Agent Card safely lets agents use consumers’ existing payment cards.
“Agent card is not a new card that you give to your agent. We actually help you use your chase card or whatever it is that you have so the agent can use it in a way that's safe.”
Listen at 53:57
Agentic commerce will become larger than early e-commerce.
“Our guess is that this is going to be larger than when e-commerce started”
Listen at 59:19
Nori’s humanoid robot can perform almost any task within its design scope.
“It's able to do almost anything.”
Listen at 1:01:01
Nori’s robot can handle 80–90% of common household tasks.
“It's able to handle it to 80 to 90% of the majority of home tasks.”
Listen at 1:01:53
Coding agents could make robot programming dramatically faster and more useful.
“if it can take under like a second for the robot to write a thousand lines of code and deploy it to the robot, it's kind of a game changer”
Listen at 1:04:54
Nori aims to manufacture about 400 robots monthly within six months.
“our goal is to basically, by the end of six months from now, to ramp up to around 400 robots per month manufactured here in San Francisco.”
Listen at 1:06:28
Nori’s claw can perform about 90% of a dexterous hand’s functions.
“you're able to do 90% of what a hand can do with a claw for now”
Listen at 1:08:15
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.