A trading firm called Wintermute, alongside Galaxy Digital, made money buying USDC on exchanges at around 99.9 cents and redeeming it near par through a bank called Erebor that was offering free stablecoin-to-cash redemptions with no fees.
Ram Ahluwalia had an early look at Erebor at a lower valuation and passed, for a reason that has nothing to do with the arbitrage: banks trade on tangible book value once they go public, and nobody enforces that discipline while a company is still private.
"I approve this message Mark Carney."
Ahluwalia runs Lumida Wealth Management, passed on the same reasoning on Anduril before its valuation tripled, and is currently sitting out a marquee AI raise most investors would fight to get into.
The full segment is covered here so you can skip it. 19 minutes of audio, 7 minutes of reading.
Here are the 6 takeaways that matter.
👤 Speaker: Ram Ahluwalia, CEO of Lumida Wealth Management
📰 Published: 16 September 2026 on the Lumida Wealth YouTube channel
🔴 YouTube | ⏱️ 19 min | ✅ Time saved: 11 min
Key Takeaways
A bank called Erebor let traders arbitrage near-par stablecoin redemptions against it Wintermute and Galaxy Digital reportedly bought USDC around 99.9 cents and redeemed it near par through Erebor's fee-free stablecoin-to-cash offer
He passed on Erebor's raise at a lower valuation over a public-market risk, not the arbitrage Private markets ignore tangible book value; public bank investors don't
He has no interest in backing another large language model or another neocloud He calls recent AI progress "linear development," not a breakthrough, even as roughly 60 LLM startups sit at unicorn valuations
Canada's new accelerated-depreciation policy is good, but he doubts it's enough So much entrepreneurial talent has left Alberta's oil patch for Texas's Permian Basin that his own diligence contacts told him "they're not here anymore"
He sees a Magnetar-style opportunity in refinancing small-cap companies loaded with debt Magnetar turned a Coreweave debt deal's equity warrants into bigger returns than the debt itself
He told a university student to skip hedge funds and prediction markets and go build at Y Combinator instead He called long-short hedge funds and prediction-market startups ideas "that made sense 30 years ago"
1. The Erebor Arbitrage
Ahluwalia opened with what he called "a funny one": a trading firm exploiting a fee-free redemption offer at a new bank.
Wintermute, along with Galaxy Digital, bought USDC on exchanges at around 99.9 cents and redeemed it near par through Erebor, a bank offering free stablecoin-to-cash redemptions with no fees. "Some trader at Wintermute got a nice bonus," he said.
Erebor is reportedly raising at an $8 billion valuation, backed by people close to Palmer Luckey, who Ahluwalia said is mostly focused on Anduril and lent his name and network — including his own WhatsApp group of contacts — to the deal.
His objection isn't the arbitrage itself; it's what happens once Erebor goes public. Private markets don't price banks on tangible book value the way public markets do, and that gap is what worried him when Lumida had an early look at the deal at a lower price.
He drew a direct parallel to why he passed on Anduril: the valuation went from $60 billion to conversations near $100 billion, while more established defense primes with more revenue — Northrop Grumman, Lockheed Martin and Raytheon — trade around $200 billion.
2. Defense Room, AI Crowded
Ahluwalia said defense stocks have sold off partly on rotation into other themes, but that shouldn't be read as a comment on near-term Pentagon spending.
He said he is watching Raytheon closely for its interceptor business, and that Pentagon spending increases of around 50% represent incremental growth for the sector rather than a zero-sum shift away from it.
On AI, he said the theme has moved past its most investable phase. "We don't need another neo cloud." He named Coreweave as the neocloud his firm did back when it was the first of its kind to go public, and said the market now has too many.
He said he has no interest in Safe Superintelligence's next funding round, despite calling its founder, Ilya Sutskever, "an extraordinary individual" who was "seminal to the transformer paper." He extended the same respect to OpenAI's Mira Murati.
His view on the technology itself: current progress is compute, data and reinforcement learning compounding along a line, not a breakthrough. He said there are now roughly 60 large-language-model startups at unicorn status, and most of them aren't making money.
3. Canada's Policy, Its Limits
The University of Waterloo trained one of the best engineers he's ever hired, which is what led him into a wider point about where Canada's talent has actually gone.
Canada's government introduced accelerated depreciation on business investment, including oil-and-gas pipelines and AI infrastructure, letting companies expense an asset's full cost in year one rather than amortizing it over time. "I approve this message Mark Carney."
He called the policy pro-investment and a genuinely good idea, then immediately questioned whether policy alone can restart what he called "animal spirits." He credited the term to John Maynard Keynes and traced the underlying idea to Joseph Schumpeter's concept of creative destruction.
Lumida was diligencing an Alberta oil-and-gas investment and reached out to contacts there to check on talent. Their answer: the people they were looking for had moved to the Permian Basin, around Midland, Texas.
He drew the same comparison he made about the Federal Reserve: officials "oversteered" by cutting rates last year while the economy was strong, and are now considering hikes with inflation still running above target. "They oversteered," he said, adding that the market is pricing a 92% chance of a rate increase at the next decision.
4. The New Western Data Trade
Ahluwalia named an oil-and-gas-adjacent holding built around a regulatory quirk: Canada's restrictions on flaring natural gas.
Thousands of productive wells sit without pipeline access and can't legally flare off the associated natural gas, so operators are forced to shut them in even though the reserves would be profitable in Texas.
New Western Data's approach is to convert that stranded gas into data-center compute using ASICs, which frees the oil for trucking and sale at around $40 a barrel. He called it a genuinely clever solution to a regulatory bottleneck.
5. Magnetar for Small Caps
Rate pressure is exposing a refinancing problem in small-cap stocks that Ahluwalia thinks is investable.
He said large companies have mostly refinanced their debt already, but many small firms in the Russell 2000 carry heavy debt loads with five-year maturities coming due that they may not be able to roll over.
His proposed playbook is a structured-solution fund that rescues over-levered small caps with audited financials and existing public-market compliance already in place. "This is what got Magnetar in business," he said, describing how the firm built a credit facility for Coreweave, took equity warrants against it, and ended up making more from those warrants than from the debt itself.
He framed it as a gap institutional capital isn't filling: private-market pricing is expensive, and larger funds like Blackstone don't chase deals this small.
6. Skip the Fund, Go to YC
A conversation with a University of Chicago sophomore turned into Ahluwalia's clearest career advice of the segment.
The student, who started a poker club sponsored by firms including Citadel and Susquehanna, asked whether he should pursue prediction markets like Polymarket and Kalshi, or join a long-short hedge fund.
Ahluwalia told him no to both. He called long-short investing an idea "that made sense 30 years ago" when Julian Robertson was building the Tiger Cub dynasty, and said prediction-market startups are chasing a trend rather than building durable skills.
"That is the best education you can get." His advice instead was to join or start a Y Combinator company: it teaches team-building, execution and shipping product faster than an MBA would, and builds a peer network that compounds over a career.
He said the advice applies even to people already in their 30s — move to San Francisco, lower your cost of living, and take the shot.
Bonus Insights
Ahluwalia described a conversation with a venture fund he has known for five or six years whose reported multiple on invested capital is 81 times, though not yet fully realized. He raised the obvious question about survivorship: in a field of thousands of VC funds, is that skill or the equivalent of flipping ten heads in a row? He said the fund had already banked real distributions as well, not just paper gains, and that it had timed an exit from the 2021 crypto bubble before describing what its second bubble call was — the transcript cuts off at that point.
On his own kids' education, he said he still funds 529 college-savings accounts but increasingly believes the highest-value path is building a company with them directly, even though a family-run business can create its own awkward dynamics.
Products, Companies & Tools Mentioned
Erebor (The bank Wintermute and Galaxy Digital allegedly arbitraged on stablecoin redemptions; Ahluwalia passed on its raise over public-market valuation risk)
Anduril (The defense-tech company he compared Erebor's valuation run-up to, and passed on for the same reason)
Raytheon, Northrop Grumman and Lockheed Martin (The established defense primes he benchmarks Anduril's valuation against)
CoreWeave (Lumida's own early neocloud investment, and the company behind Magnetar's warrant trade he cites as a model)
Safe Superintelligence (Ilya Sutskever's AI lab; he respects Sutskever but has no interest in its next funding round)
Magnetar Capital (The firm he holds up as the model for turning a small-cap debt deal into equity-like returns)
Polymarket and Kalshi (The prediction-market startups he told a student not to chase)
Y Combinator (His recommended alternative to a hedge-fund career or an MBA)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: