a16z's Size, Ben Horowitz Joins, AI needs a (Steve) Jobs, Tim Cook's Retirement, WSJ Mansion Section

Watch on YouTube ↗  |  January 10, 2026 at 03:18  |  3:09:15  |  TBPN
Speakers
Jeremie Eliahou Ontiveros — Substack author, SemiAnalysis
Alex Rampell — General Partner, a16z Apps Fund
David George — Senior Research Analyst, Baird
Ben Horowitz — Co-Founder, Andreessen Horowitz (a16z)
Jen Kha — Head of Investor Relations, a16z

Summary

TBPN covers Apple CEO succession rumors, AI narrative and techlash, Meta's nuclear power deals, AI recycling, and a16z's $15B fundraise. Guests from SemiAnalysis and a16z discuss AI data-center power constraints, gas turbines, utilities, hyperscaler buildouts, AI application investing, and bubble psychology. Main market implications include continued AI infrastructure demand, near-term reliance on natural gas for data centers, and a constructive but selective view on AI-native software and public market IPOs.

  • Apple CEO succession rumors and product roadmap debate.
  • AI narrative problem: leaders' doom quotes versus need for human-centered visionary.
  • Meta nuclear agreements with Oklo, Terrapower, Vistra; power demand mismatch expected 2027-28.
  • SemiAnalysis: AI grid queue overwhelmed; gas is near-term power solution, nuclear longer-term.
  • Hyperscaler data-center construction and leasing accelerating into 2026-27; utilities see load growth.
  • a16z raised $15B; LPs favor holding winning private companies; IPO window expected to improve.
  • AI application investing themes: green field, labor replacement, proprietary data; Toast highlighted.
  • Ben Horowitz argues AI is not a bubble because bubbles form when nobody fears one.
Ideas
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 94:25
Gas is only near-term AI power solution.
The AI power crisis is overwhelming the grid because everyone is spamming interconnection requests, so hyperscalers and labs are turning to on-site natural gas. He counts 12 manufacturers with over 400 MW of US data center gas-power orders, names GE Vernova and Siemens Energy as key players, and says for the next few quarters gas is the only solution; nuclear takes years and solar/battery is not ready. This is a long on natural gas and gas turbine suppliers.
UNG GEV ENR.DE HIGH
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 97:52
Existing power sites enable miner AI pivot.
Bitcoin miners already have power deals, substations, transformers, and energized sites, so they can retrofit and lease megawatts to AI data centers without waiting in the grid interconnection queue. This makes them valuable AI infrastructure hosts.
WGMI MED
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 102:35
Amazon construction acceleration precedes AWS revenue growth.
Amazon's data-center construction starts accelerated dramatically in 2024, and because construction leads revenue by about a year, this should show up as AWS/AI revenue acceleration, contradicting the view that Amazon is losing AI.
AMZN MED
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 106:07
Utilities benefit from accelerating data-center load growth.
AI data-center load growth is already moving the needle and will accelerate for at least two years. Every publicly traded utility he tracks is discussing data-center load growth and capacity constraints, making the utility sector a direct beneficiary of AI infrastructure buildout.
XLU HIGH
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 110:46
Bloom's fuel-cell payback offsets higher cost.
Bloom Energy's fuel cells cost more than turbines, but because AI labs pay high prices per megawatt, the payback period on new capacity is shorter. That means Bloom can expand manufacturing with less financial risk and is well positioned to supply on-site power for AI data centers.
BE MED
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 116:38
Meta redesigns data centers for AI speed.
Meta has redesigned its data centers for speed over energy efficiency, moving from two-year H-shaped builds to rectangular designs and tents, with a goal of deploying a GPU cluster in six months. Its 2.1 GW Louisiana campus and liquid-to-air sidecars show it is prioritizing rapid AI scaling.
META MED
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis 121:20
Google's networking enables multi-site AI training advantage.
Google is the only hyperscaler doing multi-data-center training at scale because it has superior networking/fiber and model architecture. This lets it stitch together many 200-500 MW sites into gigawatt-scale clusters instead of finding one huge site, giving it more site-selection flexibility.
GOOG MED
Alex Rampell General Partner, a16z Apps Fund 153:36
Toast is favorite vertical restaurant software OS.
Alex Rampell calls Toast one of his favorite businesses: it is a vertical operating system for restaurants, bundling payments, payroll, and software. AI expands the TAM because it can handle labor and cases that SMBs previously could not afford, making them software buyers.
TOST HIGH
David George Senior Research Analyst, Baird 159:52
AI-native apps show superior revenue per employee.
David George says AI-native application companies are run very differently, generating $500k to $5M revenue per employee versus about $400k for legacy software, and they face a tsunami of demand. The post-COVID founder vibe shift toward harder work is accelerating them.
AI-SECTOR MED
Ben Horowitz Co-Founder, Andreessen Horowitz (a16z) 174:59
AI not bubble; everyone fears one.
Ben Horowitz argues AI is not in a bubble because bubbles form when nobody believes it is a bubble; today everyone is debating one. The technology is working and scaling, with ChatGPT revenue going from zero in 2022 to an estimated $15B-$20B, unlike the dot-com era where valuations ran ahead of a small internet.
AI-SECTOR HIGH
Up Next
1:09:09
Jason’s Top CES Products and Takeaways | E2232
This Week in Startups · Jan 10
30:17
AI Needs a Steve Jobs, Meta’s Nuclear Powerplants, a16z raises $15B | Diet TBPN
TBPN · Jan 10
20:07
트럼프가 밀자 은행이 공포에 빠졌다… 스테이블코인이 만들 금융 혁명 "AI시대, 은행이 생존하기 어려운 이유" | 정구태 대표
815 Money Talk (815머니톡) · Jan 10
58:37
밝혀진 비리만 '1조원...?' 어떻게 계속 집권 했을까 | 강성용 서울대학교 남아시아센터장 [강성용의 남아시아 인사이드]
3PRO TV (삼프로TV) · Jan 10
6:35
Cambodia Deputy PM Pushes to Cool Thai Border Tensions
Bloomberg Markets · Jan 10

This TBPN video, published January 10, 2026, features Jeremie Eliahou Ontiveros, Alex Rampell, David George, Ben Horowitz discussing UNG, GEV, ENR.DE, WGMI, AMZN, XLU, BE, META, GOOG, TOST, AI-SECTOR. 10 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Jeremie Eliahou Ontiveros, Alex Rampell, David George, Ben Horowitz  · Tickers: UNG, GEV, ENR.DE, WGMI, AMZN, XLU, BE, META, GOOG, TOST, AI-SECTOR