Summary
High-profile FII panel moderated by Peter Diamandis with Travis Kalanick (Uber founder, CloudKitchens CEO), Jack Hidary (Sandbox AQ CEO), Eric Schmidt (former Google/Alphabet CEO/chairman), Ruth Porat (Alphabet/Google President & CIO), Ben Horowitz (a16z co-founder), Shou Zi Chew (TikTok CEO), and Jay Puri (Nvidia EVP). Topics span AI’s immediate business impact, Large Quantitative Models (LQMs), AGI timeline (Schmidt: 6-8 years to 90% expert-level across all fields), AI value chain investment, and national AI infrastructure strategy.
Key Segments
- [00:02] Travis Kalanick on CloudKitchens: autonomous food-making robots with personality; restaurants with zero/one human
- [00:04] If Travis were still Uber CEO: would have built on the attention paper, run transportation as high-frequency trading
- [00:09] Jack Hidary on Large Quantitative Models (LQMs): AI trained on physics/chemistry equations, not internet data
- [00:11] Sandbox AQ + Sanofi partnership for drug discovery using quantitative AI (announced same day)
- [00:16] Eric Schmidt: AGI in 6-8 years; single system at 80-90% of expert ability across every field by ~2030-2032
- [00:19] Schmidt on AI danger: proliferation of mid-tier models; $100M training cost as rough danger threshold; biology as most concerning vector
- [00:23] Ruth Porat: AI safety and upside are “two sides of same coin”; AlphaFold Nobel Prize; Google translating 260 languages (110 added in 6 months)
- [00:28] Ben Horowitz on value chain: bottlenecks shift (bandwidth 1999, chips 2024, power next); dev tool company 0 to $40M ARR in 3 months
- [00:30] Horowitz: token prices dropped 100-fold in 2 years but revenue still growing; GPT-2 to 3.5 jump larger than 3.5 to 4 (diminishing returns signal)
- [00:33] TikTok CEO on AI creative tools: lowering barrier from idea to content; AI content moderation as real use case
- [00:38] Nvidia EVP: AI computing advanced 1 million times vs 10x from Moore’s Law; “AI Factory” as national infrastructure
Notable Claims
- CloudKitchens has fully automated restaurants operating today with zero or one person
- Sandbox AQ + Sanofi drug discovery partnership announced at FII
- Eric Schmidt: 6-8 years to a system that is 90% of the best expert in every field
- Schmidt’s rough threshold: AI models costing <$100M to train are “probably not dangerous”
- a16z portfolio company: dev tool went from $0 to $40M revenue in 3 months (record)
- Google added 110 languages to translation in 6 months, now at 260 total (500M people served)
- Nvidia: accelerated computing advanced AI compute 1,000,000x vs general computing’s 10x over same period
- Battery technology in use today is 45 years old (Hidary)
Guests
- Travis Kalanick — Uber founder, CloudKitchens CEO
- Jack Hidary — CEO Sandbox AQ
- Eric Schmidt — Former CEO/Chairman Google/Alphabet
- Ruth Porat — President & CIO, Alphabet/Google
- Ben Horowitz — Co-founder/GP, a16z
- Shou Zi Chew — CEO, TikTok
- Jay Puri — EVP Worldwide, Nvidia
Assessment
The highest-signal episode in this batch. Seven heavyweight panelists each bringing genuinely different perspectives rather than echoing each other. Schmidt’s AGI timeline (90% expert-level across all fields by 2030-2032) is the most specific and quotable prediction. Horowitz’s bottleneck-shifting analysis (bandwidth to chips to power to cooling) is excellent strategic framing for investors. Hidary’s LQM concept — AI trained on physics equations rather than internet text — is a meaningful distinction that most AI coverage misses. Travis’s CloudKitchens reveal (fully automated restaurants operating today) is under-reported. The 6-minute-per-panelist format means nothing gets deep treatment, but the density of concrete claims makes this a strong reference document. Sponsor contamination present (Fountain Life, Viome, OneSkin) but contained in clear breaks.