2026-03-21
NVIDIA's $1 Trillion Prediction, Anthropic Beats OpenAI, Tesla vs. TSMC & The CS Job Collapse | 240
Fresh off NVIDIA's GTC 2026 and the Abundance Summit, the Moonshots crew break down Jensen Huang's trillion-dollar-by-2027 bookings claim, the viral spread of Open Claw and its enterprise cousin Nemo Claw, and why TSMC fab capacity (not demand) is now the ceiling on the entire AI industry. They cover Sam Altman's claimed 1,000x inference cost drop, Anthropic's dramatic enterprise-market-share flip against OpenAI, Elon Musk's newly announced Terafab chip-manufacturing ambitions, and a wave of nuclear-energy and robotics news (Nvidia's robo-taxi partners, Travis Kalanick's Atoms venture, the first US robotic sports league). A Fountain Life segment covers dementia prevention, and the back half ranges into UBI/UHI economics, a stark collapse in computer-science job placement rates, an AMA on mind uploading and patents, and a closing tangent on the White House registering aliens.gov.
Topics
NVIDIA GTC 2026: $1 trillion bookings and the TSMC bottleneck Compute ▶ 6:01
Open Claw / Nemo Claw and the 'organizational singularity' AI ▶ 9:20
NVIDIA's physical AI push: robo-taxis and robotics partnerships Robotics ▶ 19:16
Is NVIDIA becoming a regulated monopoly/kingmaker? Geopolitics ▶ 20:52
Orbital data centers and neutrino communication Space ▶ 25:42
Sam Altman's 1,000x inference cost drop Compute ▶ 31:42
Anthropic overtakes OpenAI in enterprise market share Economy ▶ 45:57
Elon Musk's Terafab chip manufacturing ambitions Compute ▶ 63:12
Physical Superintelligence's open-source physics AI (GPD) AI ▶ 74:04
Nuclear energy resurgence to power AI data centers Energy ▶ 84:13
Travis Kalanick's Atoms: the 'food computer' robotics venture Robotics ▶ 89:36
UBI, UHI, and the CS/knowledge-work job market collapse Economy ▶ 102:23
Predictions made
open Jensen Huang (clip): NVIDIA will book at least $1 trillion in revenue looking out through 2027.
“Right here where I stand, I see through 2027, at least $1 trillion.”
Your call:
open Alex Wissner-Gross: A new open-source AI project will repeat Open Claw's explosive growth pattern (potentially going from zero to a billion stars in minutes), with the cycle time shrinking to a few months or at most 2 years.
“maybe in a few months or at say at maximum 2 years, we were having a similar discussion... this new repo from 2027 went from zero to a billion stars in five minutes”
Your call:
open Dave Blundin: NVIDIA will generate about $350 billion in revenue this calendar year and grow another 2x once the 2-nanometer node comes online, after which it will be floored by fab capacity.
“It'll be 350 billion dollars this calendar year. And it'll grow at the max possible rate that he can get TSMC capacity... he can grow another 2x into the 2 nanometer node and then he's floored.”
Your call:
open Alex Wissner-Gross: Within a few years, improved physics will enable practical neutrino-based communication ('neutrino phones') offering ultra-low latency straight through the Earth.
“one can imagine in a few years when we have better physics having neutrino phones that just go straight through the earth”
Your call:
open Salim Ismail: The pace of inference-cost optimization (1,000x in 16 months) will continue enough that massive new data-center and energy buildouts may not actually be necessary.
“I'll make a prediction that the optimization we're doing, which is 1,000 X in 16 months, is going to keep going... in such a way that we may not need to tile the world with data centers or energy.”
Your call:
open Alex Wissner-Gross: A genuinely new post-transformer AI architecture (comparable in impact to the LSTM-to-transformer leap) will emerge soon, likely within the next year, and likely won't map well onto current NVIDIA hardware.
“it probably will it'll be soon like in the next year... it probably won't map well to the current Nvidia architecture”
Your call:
open Alex Wissner-Gross: All five major AI frontier labs will eventually become multi-trillion-dollar companies, not just one winner.
“All of all five of the major labs are going to be worth trillions and trillions of dollars... Biggest companies you've ever seen.”
Your call:
open Dave Blundin: Government jobs (and similar institutions like universities) will be the last category of jobs to be automated, because institutions will keep paying people regardless of productivity need.
“My advice is I think the last job to be automated will be government jobs.”
Your call:
open Dave Blundin: A US state will pass a China-style law requiring companies to retrain AI-displaced workers, triggering a rush by employers in other states to lay off workers before similar laws reach them.
“that'll happen in the US very soon, but it's going to roll out in one state first... everyone in all the other states is going to race to fire people before the law passes in their state, too.”
Your call:
open Alex Wissner-Gross: The White House is preparing to make a significant, non-obvious public statement on UAPs/non-human intelligence within the next few months (rumored June, July, or summer 2026).
“it sounds like the White House is preparing to say something interesting on the subject in the next few months... there are rumors of July. There are rumors of the summer.”
Your call:
Numbers that matter
- GTC 2026 drew 30,000 attendees, 2,000 speakers, 1,000 sessions
- NVIDIA has locked up ~70% of TSMC's 3-nanometer node volume
- NVIDIA's gross margin is already at 80%
- NVIDIA projected at $350 billion revenue this calendar year
- Reasoning-model cost fell about 1,000x from OpenAI's o1 (16 months prior) to GPT-5.4
- Anthropic's first-time enterprise customer share rose from 40% to 73%, while OpenAI's fell from 60% to 26%, over 3 months
- OpenAI is scaling back roughly $1.6 trillion in planned Stargate data-center spending
- Elon Musk's Terafab targets scaling from 100,000 to 1,000,000 wafer starts per month, ~70% of TSMC's current annual global output, with $100-200 billion in custom chip investment
- TSMC runs roughly 150,000 wafers/month per major process node across 3 major nodes, totaling ~500,000 wafers/month (~6 million wafers/year)
- ASML manufactures about 700 EUV lithography machines per year, with plans to possibly reach 1,000
- NVIDIA's robo-taxi partners now include BYD, Hyundai, Nissan, and Geely, representing 18 million cars built each year, joining Mercedes, Toyota, and GM
- Conservative estimates say 45% of dementia cases are entirely preventable; 1 in 4 Fountain Life members tested had advanced brain age; healthy-living interventions improved brain age by 26%
- Morgan Stanley projects a data-center power shortfall of 13 to as much as 44 gigawatts through 2028
- Meta secured 6.6 gigawatts of nuclear power for 2035 via a TerraPower partnership
- Japan's restarted reactor (TEPCO reactor #6) is projected to supply 20% of Japan's electric needs by 2040
Worth digging into
🕳️ Elon Musk's Terafab chip plant
🕳️ Physical Superintelligence's 'Get Physics Done' (GPD)
🕳️ Aliens.gov and the White House UAP disclosure timeline
🕳️ Eon Systems' whole-brain emulation of a fruit fly
🕳️ TSMC/ASML manufacturing bottleneck math
🕳️ CS job placement collapse data source