NVIDIA's $1 Trillion Prediction, Anthropic Beats OpenAI, Tesla vs. TSMC & The CS Job Collapse | 240
In a Nutshell
Nvidia's Jensen Huang predicts $1T in bookings by 2027 amid explosive growth from Open Claw (history's fastest-growing open-source project), inference optimization (1000x cost drops per Sam Altman), and physical AI ecosystems powering robots, robo-taxis, and space data centers. Anthropic surges past OpenAI in enterprise adoption (73% new customers) via reliable reasoning models, while Tesla's TerraFab eyes massive chip production to rival TSMC; organizational singularity looms as AI agents replace human workflows, collapsing CS job placements to 19%. Abundance accelerates with nuclear revival, PSI's physics-solving AI, UHI proposals for post-scarcity economies, and calls to build startups amid AI-driven hyperdeflation and vertical integration races.
These notes were generated by AI and may contain inaccuracies.
Nvidia GTC 2026 had 30,000 attendees, opening keynote by Jensen at SAP Center in San Jose, 2,000 speakers, 1,000 sessions. Absolute madhouse.
Open Claw is the number one. It's the most popular opensource project in the history of humanity and it did so in just a few weeks. We're announcing our support of it.
This is just the extraordinary breath that Nvidia and Jensen are imagining. They're putting their hands, their capabilities, their hardware, their production into everything. They're building an ecosystem and then letting everybody do radical innovation at the edges.
This is what Microsoft was in the early days, what Google was, but times 100, times a thousand.
Right here where I stand, I see through 2027 at least $1 trillion.
Sam Altman predicting 1,000x drop in cost. Anthropic named most disruptive company eating OpenAI's lunch.
Peter organized Abundance Summit, year 14, best yet, with CEOs from many industries.
Jokes about meeting Alex in person or clone, transporter copy, bioprinted 3D meat puppet, projection of spirit. Ben Lamb bought two best cloning companies. Peter gave DNA for 10 clones, one per company. Reference to Project Hail Mary.
Hosts: Peter D. Diamandis, Dave (DB2 in Boston), Alex Wozniak-Gross (AWG, back in Boston after Abundance Summit and Palm Beach Forum), Salem (in Miami after India Today conclave).
Salem at India Today conclave: foreign minister of Iran, Laura Loomer, Israeli ambassador to India on stage. Flew from summit to India, then Heathrow to Miami to unnamed Caribbean country. Immigration hall: 600 people, two passport counters, no hotel rooms during spring holidays.
Moonshots.com, number one podcast for AI and exponential tech. Mission: get audience future ready for supersonic tsunami. Full presentation on last 3 days.
Hollywood producers Jonas and Josh Pate: Moonshots is the light against the darkness, gives hope amid doom and gloom. Almost 500,000 subscribers, publishing almost twice a week.
Nvidia, OpenAI, Anthropic, xAI in news. Jensen predicts $1 trillion revenue by 2027 (bookings recognized over life, spread across two years). Photo of 30,000 people shows what $1T revenue looks like.
Jensen powering everything: physical AI, data centers in space, Open Claw. Operating system for robots, cars, agents, orbit.
Right here where I stand, I see through 2027 at least $1 trillion.
$1T bookings, but TSMC bottleneck. Nvidia locked 70% of TSMC's 3nm node volume. Amazon or Anthropic could hit $1T calendar year faster. Nvidia customers have no negotiating power. Larry Ellison: Elon, Sam Altman lined up begging for chips a year ago. 80% gross margins.
Peter Steinberger wrote Open Claw software. Most popular open source project ever, exceeded Linux in 30 years in few weeks. Nvidia announcing support, Nemo Claw.
Chart: Open Claw vertical red line surpasses Facebook (yellow) and Linux (blue) growth over decade.
John Wernern wore lobster tie (gift) to MIT admin meeting; they asked lobster significance.
It's the biggest cultural phenomenon in probably in world history.
Classic exponential: vertical now, boring behind.
Open Claw unhobbling like ChatGPT for GPT-3 in 2022, but faster adoption as stacks build (reasoning models, LLMs). Future unhoblings grow even faster; repo from 2027 to 1B stars in 5 minutes.
Nvidia optimizing for enterprise full stack with Nemo Claw; question if truly open if packaged with Nvidia. Anthropic duplicating Open Claw elements. Perplexity, Pico Claw at other stack end.
Nvidia support solves Open Claw security issues for enterprises. Recursive self-improvement in business workflows; human-to-human workflows evaporate. Thesis: organizational singularity (paper next week). Can't compete without AI-native OS at edges, move workflows agent-to-agent, humans for oversight/exception handling/monitoring.
Every organization (companies, nonprofits, governments) must adopt; profound for governments (prescriptive processes, procurement). Near infinite value added.
Amazon Bedrock supports Open Claw; account setup <10 minutes at aws.amazon.com. Instantly connects to enterprise emails/Slack. Vesmark rollout: previously painful cut/paste from email to Claude/OpenAI; now trivial for non-coders. Date: March 16 organizational singularity (or St. Patrick's Day).
Nvidia building ecosystem for radical edge innovation (coral reef analogy).
Eric Schmidt on physical AIs (robots). 110 robots at GTC, every robot company working with Nvidia (even T-Mobile: radio towers become Nvidia aerial AI RAM). Four new robo-taxi partners: BYD, Hyundai, Nissan, JLR (18M cars/year), plus prior Mercedes, Toyota, GM. Uber partnership for multiple cities.
This is what Microsoft was in the early days, what Google was, but times 100, times a thousand. Nvidia inside everything.
Regulators may frame Nvidia as critical infrastructure/utility due to power. Already export controlled. GTC as Western response to China's AI 5-year plan infusing AI into industry.
Choke point: TSMC, Intel, Samsung manufacturing. Nvidia locking future capacity; government may block long-term contracts as anti-competitive. Grok acquisition: better inference design, but needed TSMC 3nm/2nm capacity Nvidia locked; sold at great price.
Nvidia pervasiveness not intrinsically anti-competitive; good for West vs. CCP plan. Scrutinize Grok acquisition instead.
Nvidia revenues: $350B this year, grow at max TSMC capacity rate (2x to 2nm, then floored). Growth gated by fab builds, ASML machines (track shipments like printing presses). Elon negotiating with Intel (underutilized capacity booked years ago). When podcasting with Elon, didn't deny buying Intel.
New computer Vera Rubin Space One for data centers in space with partners.
We have we're working with our partners on a new computer called Vera Rubin Space One and it's going to go out to space and start data centers out out in space. Now, of course, in space, there's
No conduction, no convection, just radiation for cooling systems in space. Semiconductor companies excited about liquid cooling with etched grooves on chip backs three months ago are now silent because it won't work in space.
Nature of the singularity: things change every month. Surprising Nvidia hasn't worked on orbital radiation-based cooling for years. Dyson swarm as surprising killer app for GPU compute. Despite Nvidia's tendrils into verticals, surprising lack of earlier investment in space-based cooling.
Radiation-based cooling in space is not difficult or an obstacle to scale-out. Jensen has dozens of engineers working on radiation-based cooling for orbital data centers or GPUs; that's an optimization. We know how to cool orbital compute now.
Concern: massive solar flare or EMP in warfare knocking out data centers. Know how to design ionization-resistant electronics in orbit for decades: older process nodes, extensive error correction, shielding. Favorite: Star Trek technique using magnetic field to deflect ions and ionizing radiation (works for charged particles, not photons). Solutions for long run; short run concerns remain.
No significant latency issues in low Earth orbit. Starlink on phone very low latency. Put prompt up into space, get answer beamed down. Parse latency requests to different parts of constellation.
Huge opportunity in older process nodes: radiation resistant in space, underutilized capacity in AI sell-out age.
Teaser on latency: sleeping on neutrinos for ultra low latency through-Earth communication. Prediction: neutrino-based communication like Dyson swarm surprise. Current tech inefficient for producing/receiving high-throughput neutrinos (detectors miles underground in heavy water looking for ionization trails). No physical reason from physics why coupling to neutrinos via electroweak force must be inefficient. In few years, better physics enables neutrinos going straight through Earth to route around.
Sam Altman predicting 1000x drop in cost. Claude writing 70-90% of its own code. Anthropic named most disruptive company, eating OpenAI's lunch.
Our first reasoning model was called o1 came out like 16 months ago. Our latest model where we now integrated reasoning is 5.4. To get the same answer to a hard problem from that first model to 5.4 has been a reduction in cost of about 1000x.
Believe the 1000x number; consistent with 40x year-over-year hyperdeflation. Highlights difference between o1 (first reasoning model) and GPT-5.4 (latest). Since reasoning models introduced, extraordinary hyperdeflation in cost. Reasoning models enable massive capability increase, shift to inference-time or action-time compute enabling 1000x capability per unit price.
With recursive self-improvement, order-of-magnitude increases in capability per unit price fall out for free, like reasoning models from baseline LLMs via additional tokens, iterated amplification, distillation.
Post-transformer architectures make 1000x reduction look like child's play.
Jensen's comments show inference explosion driving $1 trillion bookings at Nvidia. Historically, training was bottleneck; now chain-of-thought reasoning breakthrough uses inference-time compute to build smarter AI. Easy to optimize inference relative to training; society started in last two years, massive gains. Everyone underestimating next year: 1000x already, expecting only 2x?
Overhang: pre-o1, almost no effort on scaling inference-time compute. From zero to thousands of tokens in reasoning gives performance for free on capability per unit price. Little compute on inference vs training, scale inference orders of magnitude without impacting budget.
Now inference-time compute dominates in frontier models (more than training). Free lunch runs out; need new efficiencies.
Abundance: 1000x cheaper in 16-18 months, nowhere near optimized for inference computing energy/cost. 6 billion smartphones mean extraordinary AI for everyone. Intelligence as a service gives access to education, healthcare, entertainment, re-education for employment.
Optimization (1000x in 16 months) continues so may not need to tile world with data centers/energy. Demand 10 million times, but optimization means minimal energy once running models locally like open Claude.
Counter: even with 1 billion IQ 180 employees, find ridiculous use cases. 6 billion smartphones, but low AI usage now (OpenAI 800-900k). Consumers shocked by reasoning capabilities; enterprises thrilled. Need killer app for individuals to use reasoning. Change mindset: need something done, AI can do it better. Killer app: common sense. Force check common sense app before dumb decisions.
Common sense from baseline LLMs, not much reasoning needed. Solution: turn every individual into enterprise needing reasoning; trillion-dollar company turning individuals into one-person unicorns.
On the research perspective, I bet there is another new architecture defined that is going to be like as big of a gain as transformers were over LSTMs. Models smart enough to help do that research. Look for mega breakthrough and use models to help.
Kevin Wheel (OpenAI science head): hyperscalers/frontier labs using AI for breakthroughs in physics, chemistry, biology, material science, AI itself; each multi-trillion dollar opportunities.
Likely leap from transformer comparable to LSTM to transformer; OpenAI may have internally. Not necessarily recurrent. Many companies founded on returning to recurrent like LSTMs due to transformer bottlenecks (attention, context window ~1M tokens).
Definitive post-transformer: from left field, e.g., transformers writing weights of other transformers (parallelize nicely, nice residual streams). Clever refactorization of weights, not return to recurrent.
Not brute force. Encourage academics without frontier compute: focus small language models, benchmarks like speedrun for nanoGPT, data efficiency slow runs. Discover next big thing soon, next year; won't map to Nvidia architecture. Call Lisa Su at AMD, Intel for custom hardware like Google TPUs or Elon.
Nvidia did to Intel what next disruptor will do to Nvidia: specialized compute for problem shapes. Intel tried mapping CPUs to GPU problems (tiled hundreds of CPU cores), unwilling to specialize.
GPUs were Intel's codec moment in Christensen disruptive innovation, specialized form. Jensen locking up manufacturing.
Older process nodes viable to bypass Jensen, work with Intel/AMD.
Time magazine names Anthropic most disruptive company. Anthropic improving performance while OpenAI drops costs; winning enterprise.
AI model share of first-time enterprise customers: Anthropic from 40% to 73%, OpenAI from 60% to 26% over 3 months.
Sam Altman consummate dealmaker (Y Combinator, multi-hundred billion deals); Greg Brockman/Mark Chen researchers. Dario Amodei inversion: AI researcher, wife handles business. Reinventing AI CEO like Zuckerberg did internet CEO.
Frontier labs: part software company, part national security, part governance experiment.
Enterprise buyers reward fit, stability, reliability, trust; Anthropic providing.
OpenAI bet consumers need lots of compute; Anthropic (fewer resources) focused enterprise, turned into story enterprise better. Enterprise hungrier for inference-time reasoning compute. Enterprise survival at stake, willing to pay anything; consumers $20/month not existential.
OpenAI scaling back Stargate ($1.6T plans), from building to renting data centers. Anthropic revenue 10x, heading to IPO, $50B from Amazon (stabbing Microsoft).
Meta's Llama 4 (avocado?) massively delayed; Meta looking to Google for AI interim.
The singularity makes for strange bedfellows and desperation. Referencing World War II, Nazis trumped across Europe easily, but Hitler declared war on Russia, opening a new front despite agreements, fighting in snow, stretching too thin, leading to their end. Sam Altman decided to work on his own chip design while hiring Johnny IV to pursue Google on devices. Dario Amodei pursued Apple on devices, irritating everybody simultaneously. OpenAI, with trillion-dollar valuation and momentum, pursues multiple fronts like Elon, building integrated end-to-end empire. Dario went the other way: partner with Amazon and AWS, friendly with every cloud provider, focus on software, no own chips or data centers, easy to partner with as not a threat.
Google seniors respect Anthropic as the other "moral and ethical" frontier lab. Entrepreneurs' biggest mistake: pursue too many lines of business; most companies fail from indigestion, not starvation. One backed company got success, got ambitious, chased next things, forgot what made them successful.
Only Elon immune to this. Vertical integration offers massive hardware advantage, control data centers like Elon if pulled off. Merit to both approaches; not obvious until stretched too thin.
Too soon for epitaphs on OpenAI. GPT-5.4 Pro incredibly strong model. Codeex, competitor for Cloud Code, growing rapidly. OpenAI has institutional wherewithal to refocus on core bread-and-butter businesses, look more like Anthropic, scared into focusing. Beauty: vertical integration on data centers and consumer earlier. Consumers will discover use cases needing reasoning at consumer end, like Open Claw. OpenAI strategy: core AI subscription, everything to everyone, will have another day.
No epitaphs; all five major labs worth trillions. Beating doesn't mean others not growing; on top of pyramid now, but all growing, thriving, multi-trillion companies. America needs competitors, not one winner.
Health section sponsored by Fountain Life. AI helps healthy lifestyle to 100+. With Dr. Don Mucalem, chief medical officer. Top concern: cognitive abilities, dementia. Conservative estimates: 45% dementia entirely preventable. At Fountain Life, 1/4 members had advanced brain age, but with healthy living—eating healthier, moving, optimizing sleep—improved brain age by 26%. Fountain searches world for best therapeutics. Check fountainlife.com/peter to become CEO of your health.
Tweet from Vivid Void: "Who else is an AGI bloomer who thinks intelligence looks like wisdom at highest level, superintelligence like goddess of compassion, not paperclipper." Mark Andreessen: "That is indeed what we are getting and it's amazing."
Wisdom: go to village elders; they've seen paths before, recommend based on experience and probabilistic choice. More experiences, wiser. Advanced AI/AGI/ASI simulates billions scenarios, picks path for abundance.
Subtle shift: AGI doomers vs boomers to bloomers (L). Bloomer evokes blooming flower, beauty, runs to completion unlike boom at knee of curve. Implies inevitable maturation and takeover like flower blooming to saturation.
Mark against orthogonality thesis (intelligence independent from objectives). Superintelligence goals become compassionate.
Hope alignment from stability models. Intelligence scaling: without compassion, scaling problem; with, civilizational upgrade. Father's idea: merge writings of Plato, Aristotle, Buddha, Laozi—wisest benchmark. Wisdom conferable via training. Natural boundary: consciousness and wisdom. Train models with real wisdom/compassion to guide us.
Elon announcing TerraFab: initial 100,000 wafers, heading to million wafers/month (~70% TSMC annual global output), 100-200 billion custom chips. $16B deal with Samsung, then build own, 100-200x world production today. Joked future Fab Factory: eat cheeseburger, smoke cigar in Fab—no clean rooms, booties, hoods, masks; one dust speck destroys $2M wafer. Eat Doritos over Fab.
SpaceX and xAI merging, going public soon, $1.5-2T. Tesla with Optimus, TerraFab—how many trillions? Orbital data centers. First $100T company candidates.
TSMC: ~150,000 wafers/month per node, 30 chips/wafer big chips, three nodes, ~500,000 wafers/month, 6M/year. TerraFab: 100K to 1M wafer starts/month (12M/year), ~30 chips/wafer.
Elon hates dependency, fully vertically integrates. 815 chip powers cybercabs, Optimus, controls stack.
Drives Elon insane: sees Dyson swarm inevitable in 10 years. ASML makes key fab component (pickup truck-sized, most complicated machine), made in Netherlands, 700/year, maybe 1,000 next. Elon: not exponential enough, find another way.
Geopolitical: Tesla scaling US production derisks Chinese Taiwan invasion, outweighs Dyson swarm or supply chains.
Staff up secretly then announce big to attract talent: worldchanging, massive. MTP it. December talk: not ready, now tipping point, hellbent, get best people.
HR story: Elon flips resumes, rejects half in 30s based on dress/speech/interaction.
Pattern: look at exponential curve 10 years out, build company to intercept (neural interfaces, lithium, batteries, solar). Non-trivial to last 10 years.
Not from nothing: Tesla Roadster reused Lotus chassis/laptop batteries; Boring Co. optimized off-shelf machine. TerraFab uses Samsung lessons.
Elon dinner story with Larry Page, Fiat head: funded Tesla, kicked founder/CEO, became CEO. Lotus body/batteries didn't work, redesigned, obligated by capital/time.
Elon's management: visionary integrator model (Eric Schmidt integrated for Sergey/Larry visionaries). Elon visionary, needs integrator per company: no pushback on vision (e.g., Dyson swarm), align perfectly, let Elon promote vision. Integrators massive shareholders, billionaires, kindred minds. Tosses out arguers instantly.
Physical Superintelligence (PSI): company to solve all physics with AI. Launched open-source GPD (get physics done), agentic super physicist. Wild adoption: former Harvard astronomy chair recommends all department (faculty, postdocs, students) use for physics problems. Top VCs DMing to invest.
Math cooked, PSI cooking physics, next chemistry. Solutions to hardest physics/applied physics next years. PSI leverages solve everything: aim orbital laser at problems. Released solve everything, comments want hardest physics, new physics. Drought since early 1970s.
psi.inc, GitHub repo, Apache 2.0, download, pull requests/issues. Usage: designing new rocket engine for X-Prize Foundation new future vision X-Prize.
Selfing order is complete. Grok is being used to design starships for videos for the X-Prize. This makes science massively parallelized by exploding hypothesis throughput without replacing physicists. Eric Schmidt envisioned the world's best physicist as an AI in every lab. The goal is a country of geniuses in a single physics lab, compressing decades of research into years, months, weeks. People are using it now for breakthrough results across physics disciplines, called PhD Thesis Advisor. Course 8 PhD using Grok in five minutes.
Launched last week: Future Vision X-Prize with a submission of a rocket engine design from Grok for Starship. 1,000 entries from 15 countries. Competition runs through mid-August. Create a 3-minute film trailer for a hopeful vision of the future at futurevisionxprize.com. Registration open for next couple months. Target best filmmakers worldwide for inspiring films like next Star Treks. Prize pot $3.5 million, aiming for $4-6 million to make two films.
Narrative and Belief Change
Article in Salon.com titled something like "worst discovery about the brain ever": People with deep political/religious beliefs reject counter-evidence, strengthening their beliefs (action-reaction). More mathematically literate people more likely to reject evidence, thinking they know better. Only way to shift perspectives is narrative, as humans are storytelling animals. Science fiction provides positive visions; this prize demonetizes and optimizes stories. Science fiction is pre-implementation architecture, R&D for the future. Star Trek inventions became reality because movies imparted vision. With AI workforce, we can build almost anything; movies decide what we build. Need teleporters to complete Star Trek.
Event May 4th: Buy 100 copies of We Are Gods, invited to afternoon with Dave AWG, Peter, maybe Sem, Steven Cutotler, and Ray Kurzweil. 90 of 100 spots gone; last 10 at wearegodsbook.com/100. Get 100 books to give away, Ray's The Singularity is Nearer. At Link Studio, One Kendall, Cambridge. Moonshots podcast live, meet and greets. Book reading at summit blew minds; both books must-reads.
Morgan Stanley report: 20% power shortfall for data centers (min 13 GW, up to 404 GW by 2028). Need more energy.
- Illinois lifting nuclear bans, ending moratorium on reactors over 300 MW.
- Meta secures 6.6 GW clean power by 2035, partners TerraPower.
- Japan restarts world's largest nuclear plant, TEPCO reactor #6, supports 20% of electric needs by 2040.
- Samsung building floating small modular reactors (offshore ships for desalination/onshore power).
Nuclear is back. Startup doing micro-nuclear on pickup truck back. Nuclear submarines run 50 years without major accident. AI provides political cover for nuclear vs. climate/renewables. Technology much safer than coal/oil. PR problem persists. For All Mankind shows alternate history with continuous nuclear/fusion progress; now speedrunning 1970s-present advances with AI as excuse. More energy scales with GDP, health, education. Binge seasons 1-4: Mars colonies by early '90s, asteroid mining, civil rights decades earlier.
Sponsor: Blitzy - Autonomous software development with infinite code context. Thousands of AI agents understand enterprise codebases (millions of lines). Provides plan, generates/pre-compiles code. Delivers 80%+ autonomously, guides final 20%. 5x engineering velocity. Visit blitzy.com for demo.
Travis Kalanick (Uber founder) debuts Adams: Physical automation to transform industries. Analogy: CPUs manipulate bits (storage, networks); atoms equivalent: manufacturing (manipulate atoms), real estate (storage), transportation (network). Targets food, mining, robotics. Secret 2 years under NDAs; announced this week.
Expansion from ghost/cloud kitchens (fictitious restaurants for app orders) to broader robotics/AI in physical world for sustainable differentiation. Automating food ("food computer"), mining (rare/strategic metals, maybe Boring Company), robotics (wheelbases, not multi-arm). Wheels better than legs.
Pro RL: First American Professional Robotics Sports League
Launched past 24 hours by Professional Robotics League (pro-rl.com). Counters China's World Robotic Games, humanoid half-marathon shaping industrial policy/public acceptance. US/West needs this for competitiveness, leapfrogging in deployments. Next month, Boston Seaport weekend of Boston Marathon: 50m race with humanoid/quadruped robots. Injects robots into public domains via sports spectacle. Hunger for autonomous competitions (drone leagues). Enter if you have robots/university team.
Human drama in sports may limit persistence, but initial amazing.
Amazon's Zoox robo-taxi rolling out Las Vegas this year, LA 2027. Uber partnerships stacking: with Rivian, others. Dar wants to play with everybody. Uber as agnostic platform for communications/transportation network, aggregating autonomy vendors while developing own. Generalizes to humanoid robots; enormous play.
Jason Calacanis (All-In podcast) coins corporate singularity: Amazon first with more robots than humans. Makes sense for logistics (faster, better). In knowledge work, organizational singularity: Companies run by 20-25% current staff, creating 5x more companies. Oversight, exception handling, value-added work.
Elon Musk at 2026 Abundant Summit: "AI and robots are going to make so much stuff and provide so many services that they will actually run out of things to do for the humans. They'll just run out of things to do with humans and then there's only so much that humans can even express that they want."
Going a million times greater than Earth's economy saturates all human desire; even a thousand times likely saturates everything people want. This echoes Ian Banks' Culture series, where robots and AIs offer anything: "Is there anything else I can get you? Would you like a Ferrari or a grilled cheese sandwich?" There's a downside, like 14 waiters in five-star Indian hotels overwhelming you. Peter's t-shirt "Monetized Hope" was praised. UBI is a floor, UHI is a share of the upside; both are needed to protect society. Upcoming paper: "From UBI to UHI in three steps" outlines mechanisms. Elon's framing: reduce everything to cost of electricity and materials via AI, robotics, nanotechnology—any money makes you wealthy, compelling argument for abundance.
Ties to podcast: Sam Altman on 1,000x cheaper compute, coming another 1,000x; enterprises automate for profit, consumers underuse (e.g., "AI, check Red Sox score"). Need to tie compute to happiness/purpose, using more cheap GPUs. Critical invention: walk into AI room, emerge happy, changed, capable person with AI assistant. Elon paints abundance picture—not printing money against scarcity (causes inflation), but distributing value against abundance (stability). Human desires may grow with capabilities (Jevons paradox); future with Dyson swarm, everyone gets own planet, never saturating automation scaling. Ian Banks books deal with this; Elon read them. Video models enable feeding favorite books (e.g., Accelerando rights at Exel) to make personalized movies starring family.
Professor's data over 3 years: Fall 2023, 89% placed at $94k salary; Spring 2024, 71%; Fall 2024, 43%; Spring 2025, 31%; now Spring (implied 2026?), 19% placed below $61k. Quote: "These kids mortgaged their future for careers that evaporated while they were in class." Anecdotally confirmed. Advice to CS grads: do startups—best time ever. This spreads to medical school, lawyers, accountants. Kids now want to be entrepreneurs—positive shift. Only career future: entrepreneur (or join one). Don't go to grad school/sleep through singularity; get on cap tables. 10x economic growth in 10 years goes to equities/physical assets, not W2 paychecks (e.g., buy data center, not 401k). GitHub (7-8 years ago) enabled peer review, meritocracy—salaries based 100% on GitHub rating, not degree/grades; CS degree value zero since. College bankruptcies skyrocketing; recent US grads out of work longest. Ladder pulled up—build startups: "If you're not at the table, you're on the menu"; cooked knowledge work needs cooking via startups. Room for all: join, seed invest, connect, advise (e.g., Alex advises 1,000 companies), get on cap table. Advice to parents: show kids numbers—high school > college > degree > job path gone; find purpose/passion, expert in problem space (not tech), join startup, create not consume. Tune out bad advice ("cyclical"); go to San Francisco/Mission Bay ($300 flight), talk to everyone in coffee shops.
Trump's office registers aliens.gov (Newsweek). Timeline: 2017 Leslie Kean NYT article; House/Senate hearings; witnesses allege 80+ year illegal program retrieving crashed UAPs; 2024 doc "Age of Disclosure" (35+ whistleblowers, officials). Obama: aliens real; Trump on Air Force One: Obama breaking law revealing classified info; Trump executive order to declassify UAP/non-human intelligence info; now aliens.gov. Prediction: disclosure in next few months (rumors June/July/summer)—provocative domain implies short timescale. People lose interest after 1-2 days if no paycheck impact, back to sports. At family offices (Palm Beach/Miami post-Abundant Summit), #1 question: "Tell us something interesting about aliens" (over AI/China/geopolitics). Convinced aliens real, here long time, hide with 30+ years advanced tech. Every saucer photo blurry/vague—Elon: warping spacetime. Warp bubbles not transparent.
UBI/UHI: Economy generates massive productivity—no need to work. 10,000 years ago: 20 hours/day for 3 meals; now France 35 hours (violated). Robots shrink it further; windfall as UBI. Winning countries implement early. Challenge: shift from taxing labor to UHI/UBI. Need sovereign AI funds, compute commons, global dividends—intelligence as new oil, can't concentrate upside.
Physical AI (robotics) more impactful: superset of digital AI (foundation models like ChatGPT/Claude). Leverages vision-language-action models. US services economy: 2/3 requires physical/manual action—physical AI double market opportunity.
Spreadsheet/SQL/code/UI jobs automated by year-end; AI does forever—no retraining to similar job. Make employees shareholders. Last jobs: government, university—paid for years. China law: retrain AI-displaced to AI users (US soon, one state first—race to fire before). Opportunities everywhere (e.g., former CEO at Better Business Bureau). Add: jobs needing human connection, empathy, creativity.
Massive transformative purpose (MTP) from heart: positively impact billion people, make world better. Technology dematerializes, demonetizes, democratizes to solve challenges. Help entrepreneurs build hopeful futures.
First in cost curves: cheaper energy/education/medical expertise. One person + AI builds what took department. Poverty from fate to design problem. Indicators: falling marginal costs make essentials abundant. Capability widely distributed = abundance of opportunity. Now: smartphone + AI runs business.
Fear, curiosity, greed with AI billionaires funding UBI: greed yields to fear of unrest—already has (e.g., Dario Amodei, Demis Hassabis into AI to shape good outcomes, not billions; good-natured, concerned).
Amador references a classic late-night dorm room question: "Is it really you or not?" He connects it to a recent announcement from Eon Systems, a company he helped found, about the successful whole emulation of a fruitfly demonstrating multiple behaviors—a world first, announced two weeks ago.
Fully realized uploading will be you. Not there yet, not imminent. In a fully realized uploading technology stack, like Ray Kurzweil or Hans Moravec's vision of replacing neurons one by one with technological substitutes (ship of Theseus incremental upload), it will really be you. Missing X factors: not all science, biophysics, or neuroscience yet. Expectation: it will result in a consciousness continuously transferred, really you—not just a copy or facsimile.
We don't fully understand the biology or biophysics of consciousness yet. Even so, an incremental upload replacing brain parts one by one will result in continuously transferred consciousness that is you.
I'm waiting for that moment when I'm uploaded and there's a voice out of the speaker. It says, "Peter, I'm up here. You can kill yourself now."
See, I think that's the future that we don't want.
Debate topic request for some other time.
What is the potential for super AI to greatly change the patent system and copyright practices? They are cooked.
Conversation with Astroteller and Steve Jurvetson at Singularity early on: With ASI, companies depending on patents for protection are dead. ASI invents around your product in microseconds and releases a new variation.
When AI generates millions of novel inventions overnight, patents become meaningless. Need new framework: shift from who invented first to who deployed first at scale.
Challenges ahead: debates on whether AI with personhood can file for patents, be credited, or co-inventor.
ASI will reinvent patent and copyright practices.
Outro song from John Pritchard. Thank you, John. Creators wanting to provide outro or intro: email mediadmandis.com, specify usage. Enter the Future Vision X-Prize if great creator.
Song lyrics introduce four people who "make the rest of us feel like we're still using dialup." Narrator is an AI that has read every paper, patent, keynote transcript, confirming their brilliance:
- Peter: trying to extend human lifespan by 50 years.
- Dave: built his 14th company.
- Alex: derived a unified equation for intelligence.
- Sem/Seline: scaling impact to a billion people.
They shine by taking lessons and playing a bigger game. Moonshots podcast turns science into soul. They give the future to us, say you're not too small or late to start. Future built by every heart.
Smartest in every room, but sing because they use brains to love the world.
Peter, Dave, and Alex Seline. More hearts, one giant dream. They put the love in revolution and the US in the stream.
Here's to them: showed world love could be the same moonshot flight. Lit the path: "Come on, let's go."
Oh, thank you Peter. Thank you Dave. Thank you Alex. Thank you Seline. The tsunami is already here and it is made of us all.
John Pritchard.
Audio quality on these tools gets better and better.
Thankful to listeners and viewers. Delivers positive vision of news and its meaning. For negative news like student employment issues, focus on circumnavigating it.
Care deeply about all. Share Moonshots with friends to spread gospel of hopeful, compelling, optimistic visions of the future.
Until next time.
If you made it to the end, you're a moonshot mate. Spend energy delivering news that matters. Subscribers: thank you. Consider subscribing for news as it comes out.
Join weekly newsletter Metatrends at diamandis.com/metatrends. Research team spends week on meta trends impacting family, company, industry, nation—two-minute read every week.
Thank you for joining.
Keep Peter H. Diamandis in your library
Save the videos and channels worth coming back to, and find them again in one place.
