1000x

Avi sits down with Illia Polosukhin at Token2049 in Singapore — co-author of “Attention Is All You Need,” the paper every modern AI model is built on, and co-founder of NEAR. Illia starts with his grandparents’ Sberbank savings book going worthless overnight and bread going up a hundredfold in a year in post-Soviet Ukraine, then walks through how transformers actually work: why recurrent models were too slow, what the attention mechanism really does, and why the aliens in Arrival are the right analogy. He left Google while the paper was being written to build a company that would let computers write code. What stopped him wasn’t the compute — it was that he couldn’t pay the students labeling his training data. That detour became NEAR.

From there: the consensus protocols they broke on camera, a standing bounty for fixing Solana, how NEAR went from blockchain to fintech app running four billion a month, why Illia says you should not give your AI agent a private key, the twenty trillion dollar agent commerce thesis, and the full story of the $3.8M exploit — including how they found the attacker, got all the money back, and got an apology on chain.

--

Follow Arthur: https://x.com/CryptoHayes
Follow Avi: https://x.com/AviFelman
Follow Jonah: https://x.com/jvb_xyz
Follow 1000x: https://x.com/1000xPod
Join the 1000x Telegram: https://t.me/thousandxpod
Try the 1000x Terminal: https://1000x.money

--

Timestamps:
00:00 Coming Up on 1000x...
01:01 Ukraine, Hyperinflation, And Science Fiction
02:59 “A Machine Should Write This Code”
07:30 Joining Google, And Inventing The Transformer
16:07 Leaving Google, And Stumbling Into Crypto
20:35 The Protocols They Broke On A Whiteboard
23:29 Chain Abstraction, And Building NEAR.com
33:15 What AI x Blockchain Actually Means
39:42 The $3.8M Exploit, And The Apology

--

Disclaimer: Nothing said on 1000x is a recommendation to buy or sell securities or tokens. This podcast is for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Avi, Jonah and our guests may hold positions in the companies, funds, or projects discussed.

The whole market de-risked into the Anthropic S-1. Then it bottomed the day the filing dropped.

Avi is back in New York after a month that ran from Token2049 in Singapore to a wedding in Iraq, and he opens on why that trade keeps working — the market as a weighing machine, the perception of reality getting so bad that there’s nowhere left to go, and the short blowout that followed. Intel at 85 back to 94, SanDisk up six percent, Nasdaq up two. Same trade as the tariffs. Same trade as Iran.

Then Jonah on the filing itself. A hundred and thirty-two billion in cash against forward compute commitments north of half a trillion, which leaves exponential growth, death, or a bailout — and he thinks he knows which. Founders holding 50.1% voting control is the detail he’s most bearish on. But the broader read is constructive: SK Hynix sold out through 2030, a DRAM shortage with a long runway, another two hundred billion of equity capital entering the system, and an AI trade still in the first or second inning. The thing to avoid is uncreditworthy neo clouds whose creditors end up subsidizing Meta’s buildout.

From there, the risk that actually matters before the midterms, which Jonah argues is a single Polymarket chart. Avi pushes back hard on the affordability narrative, on who’s really struggling, and on why housing supply is the whole ballgame. And they land somewhere counterintuitive: a Democratic administration probably can’t stop AI, but it can do almost anything it wants to crypto.

Plus: why shorting altcoins may be the cleanest hedge in the book, million dollar Bitcoin, Jason Calacanis being annoyingly right, and why giving up on nuclear after Three Mile Island was the worst own goal in modern economic history.

--

Follow Avi: https://x.com/AviFelman
Follow Jonah: https://x.com/jvb_xyz
Follow 1000x: https://x.com/1000xPod
Join the 1000x Telegram: https://t.me/thousandxpod
Try the 1000x Terminal: https://1000x.money

--

Timestamps:
00:00 Coming Up on 1000x...
00:36 Why The Market Bottomed On The S-1
07:57 “Israel Is A Third World Country”
12:50 $132B In Cash, Half A Trillion In Commitments
16:39 The Only Chart That Matters
20:37 Affordability, The DSA, And What Actually Breaks
30:22 Crypto Is The Real Political Risk, Not AI
33:27 Million Dollar Bitcoin
41:16 You Only Need One Token
44:05 Three Mile Island, And The Worst Own Goal In Modern History

--

Disclaimer: Nothing said on 1000x is a recommendation to buy or sell securities or tokens. This podcast is for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Avi, Jonah and our guests may hold positions in the companies, funds, or projects discussed.

Avi is back from two weeks on the road — Singapore, Istanbul, and Iraq — and Jonah has spent that time doing actual work on the credit problem nobody wants to look at.

They open on risk: the donut chain in Iraq pulling ten million a year because nobody else bothered to show up, Paul Tudor Jones’s “losers average losers,” why wealth is self-reinforcing, and the adverse selection problem that explains why the deal in your DMs is still open.

Then Jonah walks through CoreWeave. The bonds are in distressed territory, the CDS trades sixteen points up front, and six-year paper is at seventy-nine cents. He models the company burning more than thirty billion in free cash flow a year, every year, with all of it having to come from public markets. Zoom out and it’s roughly a trillion dollars of AI CapEx this year against a free cash flow deficit of a hundred and sixty-five billion, going to three hundred and thirty-four billion next year.

Then a full monologue on why compute futures are never going to work — no enduring benchmark, a fragmenting chip universe, physical delivery that’s far too bespoke, and incumbents who benefit from exactly the opacity a futures market would destroy. Jonah respectfully disagrees with Don Wilson on this one.

And at the end, the first time this decade Jonah has been worried about Bitcoin — and it has nothing to do with quantum.

--

Follow Avi: https://x.com/AviFelman
Follow Jonah: https://x.com/jvb_xyz
Follow 1000x: https://x.com/1000xPod
Join the 1000x Telegram: https://t.me/thousandxpod
Try the 1000x Terminal: https://1000x.money

--

Timestamps:
00:00 Coming Up on 1000x...
00:48 Singapore, Istanbul, And Iraq
02:24 Donut World: $10M A Year Selling Donuts In Iraq
04:22 Losers Average Losers
13:09 CoreWeave’s Bonds Are In Distress
21:05 A Trillion In CapEx, And Who Has To Fund It
22:49 The Shale Playbook: Why The Neo Clouds Get Eaten
30:49 Why Compute Futures Won’t Happen
40:46 Jonah’s New Bitcoin Worry — And It Isn’t Quantum

--

Disclaimer: Nothing said on 1000x is a recommendation to buy or sell securities or tokens. This podcast is for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Avi, Jonah and our guests may hold positions in the companies, funds, or projects discussed.

Avi sits down with Arthur Hayes at Token2049 in Singapore — the co-founder of BitMEX, the man who invented the perpetual swap, and now the founder of FLOP Network.

Arthur traces the whole arc: growing up in Buffalo with parents who worked for General Motors, reading a book in high school that convinced him to learn Chinese and move to Asia, and starting his first day on a trading desk the morning Lehman Brothers failed. Then getting fired from Citibank in 2013, finding Bitcoin, and building the exchange that would go on to invent the product every venue in crypto now runs on — including why the first version of the perp failed, and what the funding rate was actually designed to fix.

Then the hard part: why shutting BitMEX down was a relief, what it does to you to be the bearer of billions in customer assets, and what happens to the insurance fund.

Arthur argues the AI buildout is a global misallocation of capital — that none of the frontier labs make money, that the unit economics get worse rather than better as data centers get more expensive and compute gets cheaper, and that he puts 75% odds on the labs needing a government bailout to honor their compute commitments. Which, in his view, is the most bullish thing in the world for crypto.

Plus: why both political parties always print, why China can’t assign its losses either, Zcash and the privacy narrative, and the case for a cryptocurrency pegged directly to compute and designed for AI agents rather than people.

--

Follow Arthur: https://x.com/CryptoHayes
Follow Avi: https://x.com/AviFelman
Follow Jonah: https://x.com/jvb_xyz
Follow 1000x: https://x.com/1000xPod
Join the 1000x Telegram: https://t.me/thousandxpod
Try the 1000x Terminal: https://1000x.money

--

Timestamps:
00:00 Coming Up on 1000x...
00:39 Buffalo, General Motors, And The Book That Sent Him To Asia
04:29 His First Day On The Desk Was The Day Lehman Failed
09:14 Building BitMEX, And The 100x Leverage Unlock
15:26 Inventing The Perp
18:57 Shutting Down BitMEX, And The 30,000 BTC Question
26:16 “We’re Wasting Trillions On AI”
29:22 Why The Bailout Is Inevitable
40:47 FLOP: A Currency For AI Agents

--

Disclaimer: Nothing said on 1000x is a recommendation to buy or sell securities or tokens. This podcast is for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Avi, Jonah and our guests may hold positions in the companies, funds, or projects discussed.

Avi opens with the story of the last week — the RuneScape Gold allegations, what actually happened onchain, and a promise to the people who scammed his audience. Then he and Jonah get into why memecoin culture is a tax on your attention, and what separates a scam from a trade.

From there it's markets. Jonah makes the case that Iran is a cornered animal with nothing left to escalate with, that the Strait of Hormuz is losing its relevance in real time, and that the deflation coming out of it is enormously bullish for risk assets. Avi calls macro "astrology for men." Jonah takes that personally and defends macro as a game of common sense played over a long enough timeframe.

The back half is megatrends: how to spot one, why institutions structurally can't ride them, and why that's retail's biggest edge. Plus copper, uranium, AI drug discovery, the case against buying the index, and Jonah's term for the fastest-growing job in business — the meat proxy.

--

Follow Avi: https://x.com/AviFelman
Follow Jonah: https://x.com/jvb_xyz
Follow 1000x: https://x.com/1000xPod
Join the 1000x Telegram: https://t.me/thousandxpod
Try the 1000x Terminal: https://1000x.money

--

Timestamps:
00:00 Coming Up on 1000x...
00:59 Avi Responds To The Allegations
10:20 What Actually Counts As A Scam
12:32 Twitter Beef, Posers, And Getting Back To Builders
22:10 Iran Is Cornered: The Strait Of Hormuz Trade
26:47 "Macro Is Astrology For Men"
32:07 Megatrends, And Retail's Edge Over The Pros
44:44 Why Deflation Is Bullish
47:13 Meat Proxies: Doing What AI Can't
56:41 Done With Indices

--

Disclaimer: Nothing said on 1000x is a recommendation to buy or sell securities or tokens. This podcast is for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Avi, Jonah and our guests may hold positions in the companies, funds, or projects discussed.

#Crypto #Altcoins #StockMarket #Bitcoin #Altseason #1000xPodcast #Stocks #Macro #Trading #Investing #Ethereum #Solana #Fed #CryptoScam #Q4