The New Administration Means Major Change for Crypto, AI & Tech w/ Salim Ismail | EP #143
Summary
- Diamandis and Ismail see the tech-heavy Trump White House as a four-year catalyst for crypto, energy, space and AI-led deregulation, though their calls are explicitly predictions and hopes. Diamandis expects a “crypto-first country,” biotech liberalization, tougher food rules and “energy at all costs”; Ismail adds a crypto UBI-type scheme, temporary disaster-housing codes and geopolitical “grand bargains.”
- Cheaper launch could turn space from a billionaire contest into a new-space economy. Starship’s seventh flight paired a “poetry in motion” booster catch with an upper-stage breakup that Peter tentatively attributes to a fuel leak, while New Glenn launched within 12 hours; Diamandis expects launch costs per kilogram to fall perhaps 100-fold.
- NVIDIA is positioning itself beneath both the autonomous-machine boom and the developers building it. Cosmos gives students and startups a robotics platform, while Project DIGITS offers a $3,000 Grace Blackwell desktop supercomputer in May, described as 1,000 times more powerful than an average laptop. With possibly 100 humanoid-robot companies emerging, NVIDIA is “creating their future.”
- AI agents are already becoming a labor and margin story, not merely a model-performance story. ChatGPT reportedly grew from 100 million to 300 million weekly users in two years, while Salesforce’s Agentforce 2.0 increased productivity 30% and Salesforce was not hiring new engineers. Diamandis’s blunt call: “Every company is going to have to pivot toward agents to survive.”
- The speakers expect AI to simplify government, but Diamandis rejects OpenAI’s framing of gradual development and safety. Their proposed “minimum viable policy” would reduce thousands of overlapping regulations to the smallest set retaining 99% of the protection; Ismail places OpenAI’s caution in the context of an “AI war” among companies.
- Abundant energy and AI-generated materials form the episode’s deepest technology thesis. Solar-plus-storage, small modular fission, fusion and even coal are competing to feed rising demand, while Microsoft’s MatterGen could generate compounds for specified properties rather than screen candidates. Ismail calls it “the AlphaFold of materials science,” with upside from better batteries and CO2 removal—and a downside as extreme as “gray goo.”
- MicroStrategy remains their leveraged Bitcoin expression, but the leverage cuts both ways. The company bought 1,070 BTC for $101 million at roughly $94,000 each and reported BTC yields of 48% in Q4 and 74% for the fiscal year; from January 2024 to January 2025, they cite returns of 693% for MicroStrategy versus 144% for Bitcoin. Ismail will not repeat his own margin trade after it cost him one-third of his BTC.
- Humanoid robots may scale explosively, but Ismail treats Musk’s timetable and the social transition cautiously. Musk’s stated targets run from several thousand Optimus units in 2025 to 50,000 in 2026, 500,000 within three years and 10 billion by 2040; Ismail’s reply is “a grain of salt,” even as he agrees the trend will be transformative.
Deep dive
1. The administration could reprice technology’s political tailwinds
Ismail’s most aggressive prediction is a pair of geopolitical bargains: use saber-rattling capability to settle Russia–Ukraine and rebalance trade with China while giving China Taiwan “over a period of time.” He presents this as an opportunity to alter the global order, not a certainty.
His domestic wishlist couples crypto with a UBI-like structure—enough “to survive but not be happy,” in his formulation—with emergency building codes allowing one- or two-year temporary structures after disasters in Los Angeles, North Carolina and wherever comes next.
Diamandis expects humans back on the Moon “in a couple years” and a major Mars push with Elon in the White House. He also predicts biotech deregulation after four punishing years for its financial markets, while food regulation tightens under a “Make America Healthy Again” agenda.
Ismail cites a stark health diagnosis: America ranks roughly 74th in life expectancy and 73rd in healthspan while leading in health-care expenditure. Diamandis adds that 70% of the population is overweight or obese. Add energy expansion, AI-assisted regulation and Bitcoin reserves, and Diamandis sees a government tech-forward enough to light an “exponential rocket ride.”
2. Starship’s failure did not weaken the launch-cost thesis
Starship flight seven delivered both sides of experimental spaceflight: Mechazilla’s chopsticks caught the returning booster in what Diamandis calls “poetry in motion,” but Peter tentatively attributes the upper stage’s return in parts and pieces to a fuel leak and ignition.
Ismail’s framing: spaceflight is unusually binary because “everything has to go exactly right,” and a slight failure becomes catastrophic. Diamandis answers with SpaceX’s learning curve—the first three Falcon 1 flights failed before the fourth succeeded—and Musk’s line: “Success is uncertain, but entertainment is guaranteed.”
Within 12 hours, Blue Origin’s New Glenn also launched after roughly 12 years of development. Diamandis contrasts Bezos’s Moon-and-O’Neill-colony vision with Musk’s Mars objective and expects SpaceX and Blue Origin to displace legacy launch providers as price per kilogram falls 100-fold.
The concrete economic prize is infrastructure beyond Earth: mine carbonaceous chondrites for liquid oxygen and hydrogen to refuel in orbit, or metallic bodies for platinum, iridium, osmium and palladium. Ismail’s civilizational priority is “a backup of the biosphere” against existential asteroid risk.
3. The raw energy cost of orbit exposes launch inefficiency
Ismail’s thought experiment puts a 100-kilogram person and 150-kilogram spacesuit on an efficient winch to 200 kilometers, then accelerates them to orbital velocity. At $0.20 per kilowatt-hour, Gemini 1.5 Pro calculated an electricity bill of roughly $150.
That gap between theoretical energy cost and rocket expenditure leaves substantial room for more efficient systems. Yet Diamandis closes the comparison with gratitude rather than dismissal: “Thank you Elon, thank you Jeff.”
Ismail anchors the larger migration to October 31, 2001, the date he gives for the first astronauts’ departure for the International Space Station; since then, he says, at least one human has always remained off-planet. “It’s like the first molecules bubbling off this thing.”
4. NVIDIA is building the platform beneath autonomous machines
NVIDIA’s Cosmos platform lets high-school students, university teams and entrepreneurs develop autonomous vehicles and robots. Diamandis sees multimodal models—not batteries, motors or actuators—as the current unlock allowing robotics to move “10x, 100x faster.”
The platform dynamic matters to Ismail because NVIDIA can observe which subfunctions dominate usage, move them into the base layer and lift every developer. Its investments in robot companies also seed demand for chips if the market eventually reaches billions of machines.
Project DIGITS brings the same strategy to local AI: a Grace Blackwell desktop supercomputer, expected in May for $3,000 and described as 1,000 times more powerful than an average laptop. Once compute becomes that accessible, Ismail argues, the pressure will move to models and datasets.
Their historical contrast runs from punch-card mainframes and Macintosh 128Ks to today’s price-performance gains. Diamandis says conventional Moore’s law has been “shattered”; he reports that Elon saw 10x improvement every six months, or 100x per year.
5. Agents turn AI progress into an immediate workforce question
Altman’s reported figures put ChatGPT at 300 million weekly users, up from 100 million across two years, with agents expected to enter the workforce in 2025. Diamandis treats this as an existential corporate pivot rather than optional experimentation.
His evidence is Marc Benioff’s Salesforce: Benioff says Agentforce 2.0 lifted productivity 30%, and that Salesforce is not hiring any new engineers. “Every company is going to have to pivot toward agents to survive.”
Diamandis recounts an injured couple whose Uber lawsuit was invalidated after their 12-year-old ordered through Uber Eats and accepted terms barring litigation: “We can’t even figure out how to manage the internet.”
Household robots amplify the liability puzzle—changing a baby’s diaper, using a neighbor’s power outlet or malfunctioning in a garage. The unintended consequences, Ismail predicts, will create “quite a lot of mayhem.”
6. AI could produce “minimum viable policy”
Diamandis’s governance use case starts with inflation: instead of an economist guessing how to reduce it by 1%, an AI could inspect bank data, monetary flows and other signals, then specify five coordinated actions.
Ismail says that whatever industry one is in, there may be 5,000 regulations to navigate. Diamandis asks AI to find conflicts and produce the simplest non-conflicting rule set that still preserves 99% of the intended protection.
Ismail names that goal “MVP”—minimum viable policy. SpaceX offers Diamandis’s accountability model: anyone writing a component specification signs it and leaves contact details, so future engineers can ask why it exists when it blocks an improvement. Ismail likens that provenance to documenting software for maintainers.
Their caveat is political economy. Simplifying tax and regulatory systems could eliminate large amounts of legal, accounting and compliance work; the question is whether society is prepared for “that demonetization and the impact that’ll have.”
7. AGI’s definition keeps retreating as the systems improve
Altman says OpenAI is confident it knows how to build AGI “as we have traditionally understood it” and is beginning to aim at superintelligence. Diamandis objects that nobody should use AGI without defining it.
Candidate tests include Steve Wozniak’s coffee challenge—navigate the physical world and make coffee end to end—and assembling furniture from an IKEA box. Diamandis notes that current models already exceed him across many scientific and knowledge domains: “Why isn’t that AGI?”
Their shared answer is moving goalposts. Ismail says the test becomes whatever humans cannot yet break through, while Diamandis expects the same ambiguity around AGI that surrounded the Turing test: if it did not happen in the previous six months, it might in the next six.
Google’s Gemini 1.5 and Deep Research illustrate the competitive pressure: a $20-per-month Google system can work for 10–30 minutes and produce a research report. Ismail calls the market an “AI war” among companies. Diamandis interprets OpenAI’s emphasis on gradual deployment, safety and governance partly as moat-building.
8. Energy abundance has multiple competing routes
Altman predicts a net-gain fusion demonstration; Diamandis expects Helion, which Altman has backed, to show fusion working “soon.” Diamandis counts 37 venture-backed fusion companies and expects AI to accelerate solutions in materials and magnetic-field control.
Ismail’s initial answer is “no”: the commercial race is between stable fusion and solar-plus-storage, and sufficiently efficient solar with adequate storage may remove the need for fusion. He concedes AI can still move fusion’s technical frontier quickly.
Diamandis’s broader mix includes perovskite solar, fourth-generation small modular fission reactors and, “unfortunately,” more coal. He cites 50 years of nuclear submarines without an accident and says SMRs are safe enough that he would put one in his backyard.
India has roughly doubled energy output, China tripled or quadrupled it and America flatlined, in Diamandis’s account. Ismail expects every source eventually to approach zero cost, with Jevons’ paradox ensuring demand absorbs the capacity; he chides Diamandis for viewing the race through “a scarcity mentality.”
9. MatterGen moves materials discovery beyond trial and error
Microsoft’s MatterGen is presented as a generative system that creates compounds for requested properties rather than screening known candidates. Diamandis calls materials science technology’s “unsung hero,” underlying semiconductors, biology, energy and much else.
Ismail traces the precursor to the Materials Project around 2012–2013, which catalogued electrical, chemical and physical properties for roughly half a million compounds. A battery researcher could query for voltage and thermal characteristics instead of testing lithium-sulfur or lithium-air candidates sequentially.
Applying AI to that corpus means asking for an endpoint and letting the system generate a compound that reaches it. Ismail’s verdict: “This is the AlphaFold of materials science.”
The upside examples include more efficient solar cells, desalination catalysts and CO₂ removal; Ismail imagines an atmospheric compound that crystallizes carbon and rains it down. His counterexample is deliberately extreme: a malicious prompt seeking “the equivalent outcome of gray goo.”
10. Bitcoin leverage, humanoids and human agency close the loop
MicroStrategy bought 1,070 BTC for $101 million at about $94,000 each, reporting BTC yields of 48% in Q4 and 74% for the fiscal year. From January 2024 to January 2025, Diamandis cites a 693% MicroStrategy return against Bitcoin’s 144%.
That makes the company a leveraged Bitcoin vehicle, but Diamandis keeps the downside explicit: a falling BTC price also accelerates losses. Ismail learned that personally after borrowing during an earlier run and losing one-third of his Bitcoin; he now prefers Bitcoin-based ordinals and other structures.
A proposed $20 billion national reserve strikes Ismail as “nothing”—he argues for perhaps half a trillion—while Ismail is shocked that the crypto resurgence has not been greater. His answer is that the shift was already priced into Bitcoin’s move from roughly $50,000 to $100,000.
Robotics carries similarly asymmetric expectations. Unitree’s $6,000, 4-foot-3 G1 uses three LiDARs, while NVIDIA’s GR00T blueprint supports imitation-learning datasets. Ismail says thousands of robots could share improvements across a collective system, but questions humanoid bodies, anthropomorphism and Musk’s timetable.
11. Ismail’s father reframed death around agency
Ismail’s 97-year-old father, who had lived alone and driven until months earlier, underwent an assisted-death procedure in Canada. Instead of grief and suffering, Ismail witnessed “the most joyous and ecstatic and blissful experience I could possibly imagine.”
The family arranged farewells and played Beethoven’s Ninth Symphony. The physician told Ismail that many patients reach this state because they have agency: they can choose timing, preserve dignity and manage the final schedule.
His father, a lifelong atheist, began saying “see you on the other side” and seemed “late for a meeting.” When a palliative doctor offered comfort rather than death, he replied: “Then you’re of no use. Bring the doctor who’s going to give me the procedure.”
The elder Ismail’s final intellectual legacy was equally sharp. Responding to his son’s “Fixing Civilization” talk, he said the problem was not fixing but civilization itself: “We operate as apes and tribal clans with more and more powerful weapons…we have yet to civilize the world.”