YC Demo Day, Spoon = Bent (Again), Boring Company Raises, Apple Duo 2 Predictions

Watch on YouTube ↗  |  September 10, 2026 at 19:38  |  1:27:54  |  TBPN
Speakers
Garry Tan — CEO, Y Combinator
Om Buddhdev — Co-founder and CEO, Olam Labs
Karen Serfaty — Founder and CEO, Agent Card
Edward Gee — Founder and CEO, Exosat
John Coogan — Co-Host, TBPN
Jordi Hays — Co-Host, TBPN

Summary

The episode covers Y Combinator Demo Day, including interviews with founders in hard tech, AI data, agentic payments, humanoid robotics, satellite infrastructure, and cancer genomics. Hosts also discuss AI safety, the Bending Spoons/Miro acquisition, Apple's foldable iPhone, and a late-night US Open match. Garry Tan provides the most explicit market views, arguing AI compute demand is underpriced, legacy SaaS will be disrupted, and defense tech/cybersecurity are strong opportunities.

  • Y Combinator Demo Day features 200 companies, with 25% in hard tech.
  • Garry Tan says AI compute demand across semiconductors, data centers, and inference is massively underpriced.
  • Garry Tan predicts AI-native startups will disrupt legacy software and names Workday as vulnerable.
  • Guests pitch AI data, agentic payments, humanoid robots, sovereign satellites, and AI cancer genomics.
  • AI safety and cybersecurity risks are discussed amid agent swarms and prompt injection.
  • Bending Spoons acquires Miro at a steep discount, raising private tech valuation concerns.
  • Apple's foldable iPhone Duo is discussed for crease, aspect ratio, and sales mix.
  • A US Open tennis match ends at 3:33 a.m., sparking a scheduling debate.
Ideas
Garry Tan CEO, Y Combinator 31:55
Defense tech contracts favor small teams.
The current administration is awarding seven-figure contracts to small defense-tech teams because drone and AI warfare require faster innovation than the cost-plus defense industrial complex provides, creating a large opportunity for defense-tech startups.
Garry Tan CEO, Y Combinator 33:34
AI-native startups will wipe out legacy SaaS.
AI-native startups with 20 people each doing the work of 100 engineers will out-execute legacy public software companies that still operate the old way, leading to epic turnover and wiping out slower SaaS incumbents.
Garry Tan CEO, Y Combinator 34:52
AI compute demand massively underpriced.
Inference demand will explode 10,000-100,000x, and data centers, semiconductors, GPUs, and CPUs will see demand rise 1,000x; this massive compute buildout is not priced in by the market.
Garry Tan CEO, Y Combinator 37:34
AI agent threats drive cybersecurity demand.
As agent swarms and prompt injection become real infrastructure threats, practical cybersecurity products that can block attacks in real time will be in high demand, creating a commercial opportunity for AI-era cybersecurity.
Garry Tan CEO, Y Combinator 39:23
Workday will be outcompeted by AI-native rivals.
Workday is an example of a legacy software company with thousands of engineers who are not using agentic systems, so it will be outcompeted and wiped out by AI-native data/software companies that need only dozens of people.
Om Buddhdev Co-founder and CEO, Olam Labs 48:41
Data is the next AI bottleneck.
Compute is the current bottleneck for AGI/ASI, but as 10x more compute comes online, high-quality data becomes the key bottleneck; data companies that curate and sell training data have Nvidia-like viability because scaling laws depend on data.
Karen Serfaty Founder and CEO, Agent Card 59:19
Agentic payments will surpass early e-commerce.
Agentic payments will become a massive market larger than early e-commerce because AI agents can shop faster and more efficiently than humans, and platforms that let consumers safely use their existing cards with agents will enable this shift.
Cheap humanoids address labor shortages.
Low-cost humanoid robots like Nori can serve as an affordable wedge into the home and eventually hotels, grocery stores, and nursing homes, addressing labor shortages as LLM controls improve.
Edward Gee Founder and CEO, Exosat 69:44
Sovereign satellite demand creates opportunity.
There is a large opportunity in sovereign satellite communications for countries that want independence from both US and Chinese infrastructure; Exosat's launch-agnostic, non-US supply-chain model positions it to serve the global south and other non-aligned nations.
AI genomics improves cancer care.
AI agents can decode full genomic data and tumor biology to find therapies missed by static reports, and insurance payers are increasingly covering sequencing because it improves life expectancy and lowers costs, creating a precision-medicine opportunity.
Up Next

This TBPN video, published September 10, 2026, features Garry Tan, Om Buddhdev, Karen Serfaty, Antonio Lee, Edward Gee, Alfredo Gonzalez discussing ITA, Legacy SaaS, SMH, DTCR, CIBR, WDAY, AI training data, Agentic payments, Humanoid robotics, Satellite communications, AI cancer genomics. 10 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Garry Tan, Om Buddhdev, Karen Serfaty, Antonio Lee, Edward Gee, Alfredo Gonzalez  · Tickers: ITA, Legacy SaaS, SMH, DTCR, CIBR, WDAY, AI training data, Agentic payments, Humanoid robotics, Satellite communications, AI cancer genomics