Equity Research · Realized Position
NVIDIA
NVDA · Semiconductors · AI Compute
I faded the most crowded name in the market, the borrow ate me, and I covered at a loss without changing my mind.
Quan Pham · M’Squared Capital Research · August 2026
What the market is missing
My view was that NVIDIA was priced for a cycle with no down-leg, at the point of maximum psychology, after a roughly ten-fold run. I still think capex digestion and margin normalization eventually matter. The market simply refused to care on my timeline.
Why I was short
I was short as a defined-risk expression of a cyclical top — hyperscaler capex decelerating off an enormous base, a 75% gross margin inviting its own competition from custom silicon and a resurgent AMD, and universal ownership that left estimates no room for error. It was a trade against price and expectations, never against the long-term importance of the technology.
What has to happen
It needed a capex air-pocket, custom-silicon share gains showing up in the numbers, or a demand wobble at a single large customer. None arrived while I held it.
What would prove me wrong
What proved me wrong was the simplest thing: demand kept melting up and the guides kept blowing out, and the cost of staying short — an expensive borrow on a name that would not roll over — became the position's defining feature. I would rather step aside than finance a crowded trade indefinitely.
Valuation & exit
I covered at a realized loss of roughly 23%, inclusive of borrow and financing. The thesis is intact; the timing was wrong, and on a short that distinction still costs real money. Discipline here meant taking the loss, not defending it.
Position
Short
Realized return
−23%
Sector
Semiconductors
Status
Closed
Last updated
August 2026
This note is the opinion of M’Squared Capital and is provided for informational purposes only. It is not investment advice, nor an offer or solicitation to buy or sell any security. Price targets and entry and exit levels reflect the firm’s proprietary views and realized transactions; they are not forecasts of future results. The firm may hold, add to, reduce, or close any position at any time without notice. Past performance is not indicative of future results.