Michael Burry disclosed fresh shorts on Nvidia near 198, along with Tesla, Applied Materials and the semiconductor ETF, and called the whole AI rally a bubble. The financial media loves this story because it writes itself. As someone who watches positioning and options flow rather than narratives, let me offer a less dramatic read.

A 13F is a snapshot, not a thesis

The first thing to understand is that a 13F filing shows you where a fund was on the last day of the quarter, not where it is today. Burry is also famous for using put options, which means the notional headline number can look enormous while the actual capital at risk is a fraction of it. He was early on the housing bubble by years and early on the last few bubble calls by so much that being right eventually stopped mattering for anyone who followed him in. Being early and being wrong pay out the same way if you are the one holding the position.

What the flow is actually saying

Nvidia is already down about 23% from its all time high, so a good chunk of the froth has already come out before Burry ever showed up in the headlines. When I look at options positioning, I do not see the kind of euphoric one sided call buying that usually marks a real top. I see a stock that has been consolidating a monster move while the underlying business keeps signing deals. The Palantir partnership for classified government AI is the quiet example. Sovereign and air gapped deployments are exactly the kind of sticky, high margin recurring revenue that does not show up in a bubble narrative.

The bull case still rests on capex, not vibes

Here is the honest bear case that actually matters, and it has nothing to do with Burry. If the hyperscalers signal even a modest pause in AI capex, Nvidia and the entire chip complex are priced for perfection and will correct hard. That is the real risk, and it is a data point we can watch every earnings season rather than a personality we have to trust. Until the capex guidance rolls over, a famous short seller taking the other side is noise, not signal.

How I would treat it

I am not adding to Nvidia at these levels because the risk reward up here is not compelling after the run. I am also not selling a core position because of a filing that is already stale. If you want to respect the bear case, respect the version of it that is measurable, which is capex, and set your risk accordingly. Do not trade off a headline that is engineered to make you feel something.

Not financial advice, just how I read the tape.