The coordinate hunt on VARA AI

People have started floating the conspiracy theory that the recent selloff was a coordinated hunt on VARA AI. Going through my daily recaps, I found a few interesting things, and I’ll explain a few concepts along the way.

VARA is a textbook heavily levered AI-hardware long

Its real AUM is only $9.1 billion, but the notional value of positions reported in its 13F is $40.1 billion, which works out to more than 4x leverage. Its 3 largest positions are call options on GOOGL, AMZN, and NVDA, each around 10–12% of the book.

CoreWeave is a company that runs on borrowed money. It has issued unsecured bonds with a 9.625% coupon, on top of loans collateralized by GPUs.

A CDS is insurance on the company’s bonds. When the CDS price rises, the market is pricing in a higher probability of default. On Thursday, CoreWeave’s CDS spiked to a record 1,194.5 basis points, implying a five-year default probability above 60%. The bonds fell below 80 cents on the dollar, with yields topping 15%.

The market is effectively saying: your cost of borrowing is now high enough that it could consume the entire equity slice. Shareholders sit at the bottom of the capital stack, behind creditors, so the stock has to reprice along with the debt. CDS goes up, stock goes down.

VARA holds roughly 10.89 million shares of CRWV. At early-October prices that was somewhere north of $1 billion, or 15–20% of its real AUM. This is a high-volatility stock, so brokers typically set margin requirements high on it.

Worse, every other position in the book points the same direction.

NBIS, IREN, and APLD are neoclouds just like CoreWeave; LITE and VRT are data-center supply chain. When CRWV’s CDS blows out, all of them fall together. Nothing in the portfolio acts as a hedge, so the only way to meet a margin call is to sell stock. And the stock being sold is the least liquid names in the book, so each sale prints a new low, and each new low triggers the next margin call.

Connect the pieces and the entire chain starts with that one $1.25 billion bond. To push CoreWeave’s stock down, you don’t need to short the equity. Push the bond price down, the CDS widens, the stock falls on its own, and the margin calls arrive on their own.

So what does it take to knock that bond down? Under reasonable assumptions, pushing the price down 10 points would require selling bonds with a notional value in the tens of millions.

Note that’s notional. The actual loss incurred would be somewhere between a few hundred thousand and $2 million.

VARA’s CRWV position of about 10.89 million shares is worth $1.1–1.3 billion at $100–120, or 12–14% of its $9.1 billion real AUM. Looking at this one stock in isolation, assuming 50% initial margin and 40% maintenance margin, it would take a drop of about 17% to trigger a margin call on this position alone. 17% sounds like a lot. But for a fund running 4.4x leverage with 53.6% concentration in its top ten, a margin call hits the whole portfolio. The other names in the basket (NBIS, IREN, APLD, LITE, VRT) carry a beta of roughly 1.5 to CRWV, which means CRWV only needs to fall 3–4% for the entire portfolio to hit the trigger line.

The bondholders are mostly income-oriented accounts. Once the price breaks below 80 cents, their own risk controls force them to sell. There’s no one on the other side of that selling, so the price falls with almost no resistance.

In one sentence: liquidating a hedge fund with $9.1 billion in AUM and over $40 billion in notional exposure might actually cost only a few tens of millions of dollars.

The Situational Awareness episode in July worked through nearly the identical mechanism: options plus leverage, then liquidated by volatility. The only difference this time is the addition of one more pressure point, the CoreWeave bond.

Once you understand the mechanism, I think you would read this as a market-structure problem rather than anyone’s conspiracy. The trigger was always there, and sooner or later someone was going to pull it. Maybe someone pulled it deliberately this time, maybe it was pure coincidence.

Thursday’s US session actually fit the deliberate script. If someone was short the bonds and waiting to scoop up the stocks being force-liquidated, Thursday delivered exactly what they wanted. Examples: MU bounced on heavy volume off the 900 put wall, EWY printed its largest volume since August at 150, while the CDS hit a record high the same day.

The stocks got picked up cheap; the bonds kept getting pressed as the tool.

¯\_(ツ)_/¯

P.S. I hope you bought some LITE and COHR with me.