2026-01-02

AI Investor Panel: Where Smart Money Is Actually Going in AI | EP 219

Recorded live on stage at a Moonshots AI mini-summit in Saudi Arabia, Peter Diamandis moderates a panel on how the global AI buildout gets funded, with Dave Blundin (Link Exponential Ventures, MIT/Harvard-focused VC), Bonnie Chan (CEO of Hong Kong Exchange and Clearing), and Anjney Midha (a16z partner, Anthropic backer, Mistral board member). The panel agrees capital demand for AI vastly outstrips traditional venture capacity, forcing in corporate strategics, sovereign capital, and public markets to fill the gap. They dig into the shift from raw compute buildout to a scarcer resource (foundation-model tokens), identify energy/power availability as the real bottleneck on scaling, and flag risks including speculative peripheral bets (fusion, robotics, quantum), an AI-driven wealth-concentration backlash, and overheated retail-investor exposure to AI valuations. The conversation closes on a call for pension and sovereign wealth funds to get public capital exposed to frontier AI before public resentment over unshared wealth creation deepens.

▶ Watch on YouTube

Topics

AI capital demand is insatiable and reshaping venture capital Economy ▶ 4:04
Anjney Midha says essentially all available capital is flowing into AI and it's still not enough; a16z has restructured so every vertical fund (infrastructure, applications, healthcare) is effectively now an AI fund.
Corporate strategics and sovereign capital fill the VC funding gap Economy ▶ 8:59
Dave Blundin notes US venture capital (~$200B/year) is dwarfed by AI's capital needs (projected $3B/day), so Nvidia, hyperscalers, and other corporate strategics are stepping in to fund AI companies directly on cap tables.
Hong Kong IPO market as an AI funding channel Economy ▶ 7:39
Bonnie Chan describes Hong Kong Exchange leading the world in IPOs this year, with roughly half of completed and pipeline deals tied to AI, and a growing base of sophisticated retail ('protel') investors.
The capital stack: cash to GPUs to tokens Compute ▶ 11:37
Anjney Midha describes AI's capital conversion chain — raw cash buys GPUs, GPUs are converted into foundation-model tokens — with high-quality tokens now the scarcest resource, scarcer than GPUs, which are scarcer than cash.
Energy as the binding constraint on AI infrastructure Energy ▶ 12:17
Panelists agree infrastructure demand is accelerating uncapped, but data centers are bottlenecked by power density, permitting, and grid capacity rather than chip supply — new Blackwell chips are ready before the energy and cabling are.
China's energy and manufacturing advantage in AI buildout Geopolitics ▶ 14:57
Bonnie Chan argues China has an edge in AI infrastructure due to abundant renewable generation in its west/northwest regions, an advanced grid able to disperse that power nationally, and a dominant manufacturing base for embedding AI into production.
AI accelerating drug discovery and data-intensive sectors Biotech ▶ 16:19
Bonnie Chan highlights growing IPO interest in AI-driven drug discovery companies, where embedding AI into clinical trial and sample-selection workflows can dramatically speed up the traditionally data-intensive process.
MIT/Harvard vertical AI startups and near-100% hit rates AI ▶ 17:32
Dave Blundin says the number of AI startups coming out of MIT and Harvard has more than quadrupled, with vertical-use-case companies (as opposed to foundation-model builders) showing a success rate near 100% because use cases vastly outnumber available talent.
Compressed valuation timelines and a new class of young billionaires Economy ▶ 19:13
Blundin describes vertical AI startups going from ~$20-30M entry valuations to unicorn status within two years, producing 23-24-year-old founders; he can now name eight portfolio founders under 30 who became billionaires, versus three or four in his entire prior career.
Public wealth-creation gap and the risk of social backlash Geopolitics ▶ 21:53
Anjney Midha argues frontier AI wealth (e.g., Anthropic's rise to $183B in 48 months) is locked inside private capital and a small talent pool, with the public shut out; he warns of coming unrest as AI displaces IT-services jobs and cites death threats already faced by Silicon Valley leaders.
Disentangling AI-caused layoffs from prior over-hiring correction Economy ▶ 24:10
Midha and Blundin argue many current tech layoffs are a correction of 2010-2020 over-hiring and pandemic-era easy money rather than genuine AI displacement, but expect AI to get blamed for both regardless.
Risk of speculative 'peripheral' AI-adjacent investments Energy ▶ 26:04
Dave Blundin warns that capital flowing into AI is spilling into unproven, capital-intensive adjacent bets (fusion energy, robotics, quantum computing) that could fail and dent investor confidence in AI broadly, echoing the 2000-2001 dot-com crash.

Predictions made

open Peter Diamandis: US AI capital deployment will grow from about $1 billion a day today to $3 billion a day by 2030, and will likely exceed that pace.
EP #? · · due: 2030 · ▶ watch
“I expect it's going to blow through that.”
Your call:
open Anjney Midha: AI infrastructure buildout demand will keep accelerating, but progress will eventually hit a hard ceiling set by insufficient electricity supply rather than chip availability or capital.
EP #? · · due: unspecified · ▶ watch
“We just don't have enough electricity to power the chips.”
Your call:
open Anjney Midha: As AI tokenizes large portions of IT-services work in economies like India, it will cause serious near-term economic transition pain and social/political backlash that governments and institutions are not prepared for.
EP #? · · due: unspecified / near-term · ▶ watch
“I don't think governments are doing enough to realize how dire it's about to get when 30% of your IT services GDP sector gets vaporized by tokens.”
Your call:
open Dave Blundin: Some capital-intensive AI-adjacent bets, such as fusion energy, robotics, or quantum computing, will fail and turn into losses, risking a broader loss of investor confidence in AI similar to the 2000-2001 dot-com crash.
EP #? · · due: unspecified / next several years · ▶ watch
“Some of them are going to consume a ton of money and turn into losses, and that may scare off the entire investment community.”
Your call:
open Bonnie Chan: Current AI-related public market valuations may already be near a peak, and retail investors entering now risk being the last capital in before a broader correction or collapse.
EP #? · · due: unspecified · ▶ watch
“They could be the last one ... being the last ones in at the party before the whole thing collapses.”
Your call:

Numbers that matter

Worth digging into

🕳️ Mercor's valuation trajectory ($30M to $10B in ~2 years)
Cited as an extreme, near-unprecedented example of how fast AI-era startup valuations are compounding, making it a useful bellwether for whether current pricing is rational or bubble-driven.
🕳️ The US 'AI Action Plan' and data center permitting reform
Anjney Midha frames this federal plan (introduced roughly two months before this episode) as the key lever that could either unblock or continue to bottleneck the energy buildout AI infrastructure needs.
🕳️ China's grid-dispersal strategy for AI data centers
Bonnie Chan claims China's ability to move renewable power generated in its resource-rich west/northwest to data centers nationwide is a structural competitive advantage for AI buildout.
🕳️ Anthropic's seed round funding story
The claim that 21 of 22 traditional VC introductions passed on Anthropic's seed round is a striking data point about how badly mainstream Sand Hill Road capital misjudged frontier AI early on.
🕳️ Sovereign wealth and pension fund exposure to frontier AI
Midha explicitly argues institutions that steward public capital need to get exposure to AI cap tables before public resentment over unshared wealth creation boils over — a concrete, checkable claim about capital allocation.