Training Physical AI on Video Game Clips

Chamath Palihapitiya · Chamath Palihapitiya · June 28, 2026 at 16:14 · ⏱ 4 min read  | Read on Substack ↗
Summary
The newsletter reports three major developments: General Intuition raised $320M to train physical AI using video game clips, OpenAI launched its custom inference chip Jalapeño built with Broadcom, and Meta invested $900M in Indian fintech CRED to revive WhatsApp Pay. For markets, these stories highlight increasing specialization in AI hardware (ASICs, custom silicon) and the strategic push by Big Tech into payments in high-growth markets like India.
  • General Intuition raised $320M at a $2.3B valuation from Khosla, General Catalyst, Bezos, and Schmidt to train AI on video game recordings from Medal’s 17M users.
  • OpenAI and Broadcom unveiled Jalapeño, an ASIC for inference, built in nine months with OpenAI’s own models aiding design; initial deployment by end of 2026.
  • IBM announced the world’s first sub-1-nanometer transistor chip using nanostack architecture, packing 100B transistors on a fingernail-sized chip, with production expected in five years.
  • Meta invested $900M for ~20% of CRED at a $4.5B valuation, hiring its founder to run WhatsApp worldwide; WhatsApp Pay handles only 0.65% of India’s 23.2B monthly UPI transactions.
  • India’s UPI 30% market share cap deadline was pushed to December 2026, potentially opening room for WhatsApp Pay if current leaders PhonePe (46%) and Google Pay (33%) are forced to cede share.
  • Meta has spent over $6.6B over six years trying to grow WhatsApp Pay in India, with the new CRED partnership representing its largest bet yet.
Read time 4 min
Length 4,846 chars
Category finance
Ideas
Chamath Palihapitiya CEO, Social Capital
Broadcom co-developed OpenAI's custom inference chip Jalapeño, an ASIC that went from design to production in nine months. This reinforces Broadcom's role in custom silicon for top AI labs, expanding
Broadcom co-developed OpenAI's custom inference chip Jalapeño, an ASIC that went from design to production in nine months. This reinforces Broadcom's role in custom silicon for top AI labs, expanding its revenue opportunity beyond networking. Risk: Dependence on a single large customer (OpenAI) and potential for margin compression if custom chip volumes ramp rapidly.
Chamath Palihapitiya CEO, Social Capital
Celestica is named as the manufacturer of OpenAI's Jalapeño chip, alongside Broadcom. This is a high-profile contract win that could lead to additional custom AI chip manufacturing deals, boosting Cel
Celestica is named as the manufacturer of OpenAI's Jalapeño chip, alongside Broadcom. This is a high-profile contract win that could lead to additional custom AI chip manufacturing deals, boosting Celestica's semiconductor services revenue. Risk: Manufacturing execution risk and the possibility that subsequent generations are insourced by OpenAI or shifted to another partner.
Chamath Palihapitiya CEO, Social Capital
IBM announced a breakthrough sub-1-nanometer transistor chip with nanostack architecture and 100B transistors, plus 40% SRAM scaling improvement. While production is 5 years out, the technology leader
IBM announced a breakthrough sub-1-nanometer transistor chip with nanostack architecture and 100B transistors, plus 40% SRAM scaling improvement. While production is 5 years out, the technology leadership enhances IBM's semiconductor R&D credibility and potential future foundry partnerships. Risk: Long time to commercialization; competing architectures (e.g., GAA from TSMC/Samsung) may leapfrog IBM's design before it reaches production.
Chamath Palihapitiya CEO, Social Capital
Meta invested $900M for a 20% stake in CRED and installed its founder as WhatsApp head, targeting India's UPI payments market. With WhatsApp Pay at just 0.65% share and a potential cap on current lead
Meta invested $900M for a 20% stake in CRED and installed its founder as WhatsApp head, targeting India's UPI payments market. With WhatsApp Pay at just 0.65% share and a potential cap on current leaders by Dec 2026, Meta is positioning for a massive TAM expansion in digital payments. Risk: Regulatory uncertainty around the UPI cap enforcement; Meta has already spent $6.6B with little traction, and the new strategy may face integration or user adoption challenges.
More from Chamath Palihapitiya

This newsletter, published June 28, 2026, features Chamath Palihapitiya discussing AVGO, CLS, IBM, META. 4 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Chamath Palihapitiya  · Tickers: AVGO, CLS, IBM, META