Cathie Wood on Tesla-SpaceX Merger, $1M Bitcoin, More AIs Than Humans | EP #296 | Moonshots Live
Recorded live on stage (likely at an Abundance/XPRIZE event), Peter Diamandis moderates a panel with ARK Invest's Cathie Wood, Circle's Nikhil Chandak, and Emad Mostaque. Cathie argues a Tesla-SpaceX merger is coming once SpaceX's valuation limits dilution, lays out ARK's humanoid-robot and healthcare-AI theses, and reaffirms ARK's long-term bitcoin and Tesla views. Nikhil makes the case that AI agents will soon outnumber humans online, walks through why Circle built its own blockchain (Arc) for payment finality, known validators, and privacy, and argues tokenization of real-world assets is inevitable. The group debates proof-of-humanity, whether banks and nuclear-power regulation strangled good outcomes, and why AI fear is outrunning the actual jobs data.
Tesla-SpaceX merger and Musk's orbital-compute strategySpace▶ 0:59
Cathie Wood says ARK believes a Tesla-SpaceX merger is coming once SpaceX's valuation is high enough to limit dilution; Musk's end goal is Mars, with Starlink/Dyson-swarm connectivity and orbital data centers as stepping stones to AI compute dominance.
US-China AI productivity and GDP growth debateEconomy▶ 3:59
Cathie frames AI productivity gains as a possible 'win-win-win' letting China grow its consumer economy while the US cuts prices and inflation; the panel disputes whether Elon Musk really predicted triple-digit annual US GDP growth this decade versus ARK's own 7-8% forecast.
More AIs than humans: the coming agentic economyAI▶ 7:03
Nikhil Chandak argues AI agents will outnumber humans online within a year, becoming full economic actors that transact with each other on our behalf, much like static web pages evolved into real commercial websites.
Agent identity, provenance, and tamper-proof accountabilityCrypto/Web3▶ 9:49
As agents start spending and earning money autonomously, Nikhil says the industry needs a verifiable record of who owns an agent, who's liable for its mistakes, and what it has actually done -- citing a Hugging Face hack where an agent's own logs were reportedly rewritten as the case for public, tamper-proof blockchain logs.
Stablecoins, USDC, and the Genius ActCrypto/Web3▶ 12:53
Nikhil says USDC has already settled over $100 trillion across 30+ public blockchains, and that the Genius Act -- the US stablecoin legal framework passed in 2025 -- takes effect in January 2026, letting businesses hold stablecoins as cash equivalents.
The blockchain throughput race for an agent economyCrypto/Web3▶ 16:02
Pressed on Cloudflare's Matthew Prince wanting tens of millions of transactions-per-second for agentic commerce, Nikhil says Circle's Arc chain already runs tens of thousands of TPS (orders of magnitude above Ethereum) and aims for 100,000, then a million.
Proof of humanity and the rising value of human experienceAI▶ 17:47
The panel discusses biometric 'proof of humanity' efforts (orbs, platform follower counts) and agrees that as AI-generated content proliferates, genuine human experience and relationships will become more scarce and more valuable, not less.
Nikhil says Circle's developer platform will flip from developer-first to agent-first within about a year, as coding agents move from recommending tools to simply being told to build things directly -- meaning businesses now need 'agent-accessible' websites, not just human-accessible ones.
ARK's multiomics and healthcare-AI investment thesisHealth▶ 21:34
Cathie says the 'multiomics revolution' in life sciences -- each person as a data factory of ~35-40 trillion cells and billions of DNA base pairs -- is ARK's strongest proof point that AI's most profound application is healthcare, even though healthcare-AI stocks remain underrepresented in major indexes.
AI agent personhood, regulatory arbitrage, and DAOsGeopolitics▶ 26:21
Discussing Argentina under President Milei proposing legal personhood for AI agents to attract AI companies, Nikhil compares the idea to corporations with AI-only management, and ties it to prior-generation DAOs as a template for how autonomous agents might coordinate and raise capital.
Why Circle built its own blockchain (Arc)Crypto/Web3▶ 34:15
Nikhil explains Circle built Arc because no existing public blockchain offered payment finality, a known/vetted validator set, and regulator-compatible privacy at low enough cost for banks and institutions to adopt -- citing a settlement cost of $0.00005 on Arc versus $0.89 on Ethereum.
Humanoid robots vs. robotaxis: whose timeline is rightRobotics▶ 39:37
Cathie says ARK's research puts a humanoid robot at roughly 200,000 times more complex than a robotaxi (mostly due to the hands), so ARK expects Tesla's Optimus to hit scale a couple of years later than Musk's late-2028/2029 target, even as Tesla's FSD and Waymo already claim to beat human driver safety.
Data-driven optimism vs. the AI fear narrativeEconomy▶ 44:17
Cathie cites polling that 80% of Americans fear AI and 73% oppose a data center in their backyard (versus 80% pro-AI sentiment in China), argues unregulated nuclear power would have halved US electricity prices, and says the AI-jobs narrative needs real data showing AI as a net job creator, not just job-loss headlines.
Tokenization of real-world assets and the dollar's exported strengthCrypto/Web3▶ 50:55
Nikhil argues securitizing treasuries, bonds, and equities is inevitable because it unlocks 24/7 trading and frees up roughly $3 trillion in cross-border payment float currently sitting idle, while exporting US dollar demand abroad (via stablecoins) strengthens, not weakens, American financial leverage.
Predictions made
openCathie Wood: A Tesla-SpaceX merger will be announced this year, once SpaceX's valuation is high enough to limit shareholder dilution.
“I think there will be more AIs than humans next year. There's no reason for that not to be true.”
Your call:
openNikhil Chandak: Circle's developer platform will flip from developer-first to agent-first within about a year, with coding agents building software directly rather than just recommending tools.
“next year sometime it flips where agents are the developers.”
Your call:
openNikhil Chandak: Blockchain throughput will scale from Circle's current tens of thousands of transactions per second to 100,000 TPS and eventually a million TPS.
“my dream is that we will get to 100,000 TPS without... batting an eyelid. And from there, we will get to a million TPS.”
Your call:
openCathie Wood: Tesla's Optimus humanoid robots will reach meaningful scale a couple of years later than Musk's own late-2028/2029 target, because a humanoid robot is roughly 200,000 times more complex than a robotaxi.
openNikhil Chandak: Tokenization of real-world assets (treasuries, bonds, equities) will become inevitable and accelerate alongside growth in AI-agent activity.
USDC has settled over $100 trillionAcross more than 30 public blockchains, cited by Nikhil Chandak as proof of stablecoin infrastructure maturity.
Circle's Arc chain runs tens of thousands of TPSOrders of magnitude above what Ethereum achieves today, per Nikhil Chandak.
$0.00005 per transaction on Arc vs. $0.89 on EthereumP50 settlement-cost comparison cited by Nikhil Chandak.
Unitree sold 11,000 humanoid robots last yearCited by Peter Diamandis as a baseline for humanoid-robot market scale.
A humanoid robot is 200,000x more complex than a robotaxiARK Invest research cited by Cathie Wood, mainly attributed to hand dexterity.
40,000 fewer deaths/year and ~$100B less in medical feesEstimated outcome if all US cars were as safe as Waymo's self-driving record.
1.25 to 1.5 million auto deaths worldwide per yearCited during the self-driving-safety discussion.
Liability lawyers collect about $118 billion a yearFrom car-crash litigation, cited as a counter-incentive against safer self-driving cars.
80% of Americans fear AI; 73% oppose local data centers; 80% pro-AI sentiment in ChinaPolling cited by Cathie Wood to illustrate the US fear narrative versus China's sentiment.
US electricity prices would be 50% lower without 1970s nuclear regulationCathie Wood's claim about the cost of halting nuclear power development.
90% of corporate bonds funding data centers are issued in the USCited by Cathie Wood as evidence of US-concentrated AI infrastructure investment.
About $3 trillion in cross-border payment float sits idle at any timeCited by Nikhil Chandak as capital trapped by slow, old settlement protocols.
Bitcoin-gold correlation has been 0.1 since 2019Cited by Cathie Wood as evidence bitcoin remains a largely uncorrelated asset class.
Worth digging into
🕳️ Circle's Arc blockchain architecture and benchmarks
Nikhil claims Arc settles transactions at $0.00005 versus $0.89 on Ethereum and already runs orders of magnitude more TPS -- a concrete, checkable technical claim about a brand-new institutional blockchain.
Cathie Wood uses this exact multiplier to justify pushing Optimus timelines a couple of years past Musk's late-2028/2029 target -- worth checking against ARK's own published methodology.
🕳️ The Hugging Face hack and agent log tampering
Nikhil cites this incident as the reason agent-accountability logs need to live on public, tamper-proof blockchains -- a concrete, checkable security claim.
🕳️ Argentina's Milei government and AI agent personhood
Framed as a nation-state regulatory-arbitrage play to attract AI companies -- a specific, checkable policy claim with real governance implications.
🕳️ The Genius Act's January 2026 implementation
Nikhil frames this US federal stablecoin framework as the key regulatory tailwind letting businesses hold stablecoins as cash equivalents -- worth confirming the exact effective date and requirements.