Wall Street Says Buy NVIDIA. Our Model Isn't Convinced.

NVIDIA did the thing it almost always does: it beat. Second-quarter revenue came in at $96.22 billion against a $92.27 billion consensus, a 4.3% surprise, and earnings landed at $2.22 a share versus the $2.09 estimate, ahead by 6.2%. Revenue more than doubled from a year earlier. By the standards of a company that has spent two years detonating expectations this was a clean quarter rather than a shocking one, but the direction is unchanged: demand is still running hot.

The sell side is unbothered by the clean-rather-than-shocking framing. Of the 62 ratings on the stock, 59 are Buys, against two Holds and one Sell. The average 12-month price target sits at $328.73, about 43% above the recent $230.36 price, with a low of $265 and a high of $515. That spread is worth pausing on: it is not a room disagreeing about direction, it is a room disagreeing about magnitude, with the bulls pricing in more than double the low case. Wide dispersion like that is typical of a stock where the story is settled but the size of the eventual market is not.

Our own read is where things part company with the crowd. AlgovestIQ's model scores NVDA 64 out of 100 - neutral - and tags it a Relative Leader / Watch with a strong relative position and an AIQ Edge of 9/10. Neutral here is not a bearish call and not a prediction of a fall; it is the model saying the risk and reward have evened out after a very large run.

The factor breakdown underneath that 64 explains the disagreement better than any single sentence could. Quality scores 97, about as high as the model goes, reflecting NVIDIA's margins, returns and balance sheet. Momentum is a healthy 68. But Value sits at 35, the model's blunt way of saying the stock is expensive, and Risk Resilience is a middling 47. So the neutral rating is not skepticism about the business. The quality read is nearly flawless. It is the price and risk factors offsetting that quality. In plain terms: a great company, trading like everyone already knows it.

The price chart tells the same story in a different language. At $230.36 the stock trades well above its long-term trend near 199.74, after climbing from the low $100s a year ago, exactly the kind of extended run that drags a Value factor to 35 and keeps the weekly rating neutral even while momentum stays firm.

What management actually said

The reason a neutral rating has not tipped negative is the demand backdrop, and on the call management leaned in hard. The framing throughout was that growth is capped by supply, not demand: customer forecasts point to demand roughly doubling next year, but NVIDIA expects about 70% revenue growth in fiscal 2028 because it cannot build fast enough. Guidance for the current quarter is $108 billion, plus or minus 2%, roughly 12% sequential growth over the $96.22 billion just reported.

The engine is still the data center, which contributed roughly $89 billion, up 18% quarter over quarter, about 92% of total revenue. Within that, management highlighted ACIE - NVIDIA's call grouping for NeoCloud, industrial, enterprise, and related sovereign and regional demand - at $40 billion in revenue, up 25% sequentially and 138% year over year. The term is NVIDIA's own and management used it a little loosely on the call, so read it as call terminology for the non-hyperscaler base rather than a fixed reporting segment.

Management also walked through the economics it sees per gigawatt of AI-factory capacity, rising from roughly $18 billion with Hopper to $25 billion with Grace Blackwell to $40 billion with Vera Rubin - a different $40 billion from the ACIE figure, and a tidy illustration of why each generation matters more than the last.

The candor cut both ways, which is what made the call unusual. Management was direct about the constraints: memory pricing is expected to push gross margin down to 71-72% in the fourth quarter before recovering to 72-73% in fiscal 2028, and the forward outlook excludes China data-center compute revenue entirely on geopolitical uncertainty. Neither was glossed over.

That margin path is the line to watch, and it is where the proprietary data adds something. NVIDIA's gross margin today, per AlgovestIQ's fundamentals as of September 4, is 74.67%, so the guidance implies real, if modest, compression from here, driven by memory costs rather than pricing weakness. On a business earning about 75% gross margins there is plenty of room before the story changes, but the first derivative is what the market trades, and it points down for a couple of quarters.

One more piece of context in NVIDIA's favor: on AlgovestIQ's catalysts page the company has beaten estimates in all eight of its last eight reported quarters. A perfect beat rate does not guarantee a ninth, but it tells you guidance has been set conservatively, worth remembering before reading much into any single only 4% top-line surprise.

The bottom line

Strip it down and NVIDIA is a company beating on schedule, guiding to about 12% sequential growth, sitting on an elite quality profile, and facing more demand than it can supply, with a stock the Street thinks is worth 43% more and our model thinks is fairly valued after a big move. The disagreement is the signal. The quality is not in question; the price is.

If you already own it, nothing here argues for selling a compounder mid-story. If you are hunting an entry, a Value factor of 35 and a guided margin dip are the model's reminder that great company and great price are not the same sentence, and that memory costs are the thing most likely to decide which way the next print breaks.