Nvidia shares are surging as much as 6 percent on a hyper-bullish report from Evercore ISI senior equity research analyst C.J. Muse. But it isn’t immediately clear that the report presents much in the way of new information.

Nvidia’s recent run has been downright spectacular. Shares of the semiconductor company have nearly tripled in value over the past year, and the stock hit an all-time high of $179.79 in midday trading Friday.

But Muse’s Friday morning report takes it to the next level, increasing the firm’s price target from $180 to $250, blowing Wall Street’s previously highest target of $200 out of the water.

In the first sentence of the note, Muse provides some useful context for the call: “We hosted NVDA mgmt. (CEO, CFO and IR) on the road with investors over the last few days in Boston and New York.”

“As for major takeaways, our sense is management believes that investors still severely underestimates the impact of AI and the size of the potential market,” he continued. “We are only at the cusp of AI’s growth potential and NVIDIA is creating THE AI computing industry standard.”

It may indeed be useful for investors to know that Muse sniffed out bullishness on the part of the company’s own management while Muse was providing Evercore ISI clients with access to them. But that leaves open the question of how the stock will manage to stage another 48 percent rally from its Thursday $169.40 close.

Ultimately, it comes down to earnings. Without explicitly making a forecast to this effect, Muse writes in three places that the company has a “clear path” to earning $10 per share. This is a striking call; the company earned $3.63 per share over the past year, and Muse’s official forecast only goes out to 2018, in which he foresees the company earning $3.80.

Yet Muse writes that “the tremendous growth in opportunities led by AI in the Data Center and Automotive verticals supports a clear path to $10 in earnings power over the next 3-5 years … with sustained growth thereafter.”

Muse then multiples this soft $10 earnings estimate by 25, which he believes is the multiple at which the stock will trade at that time, and voila: $250.

Often this kind of analysis will involve an intermediary step, whereby an analyst will “discount” this expected price back to the present by dividing it by a number accounting for the amount of risk and the number of years until the outcome in the numerator is achieved. (For instance, let’s say an estimate with this amount of risk will pay out 10 percent per year if it proves correct. In that case, $250 would be divided by 1 + 0.1 to the fifth power, or 1.61, leaving a price target of $155.)

Muse simply skips this step, leaving him free to stop at $250 and declare it his price target. He explains: “some may push back on our not discounting back to 2018 from the 2020-2022 timeframe, and this view clearly has some merit. But considering the first mover advantage NVDA has in AI coupled with the CUDA ecosystem that is becoming the defacto standard for AI globally, we suspect that the company’s multiple will be elevated for some time as growth investors will be willing to project beyond 2020-2022.”

The call, then, appears to assume demand for the stock from a particular type of investor. It leaves open the question of what math these investors themselves will be doing that leaves them free to ignore the risk that arises naturally when lofty expectations meet long time frames.

This isn’t the first time that Nvidia has inspired analysts to take a novel tact. When he raised his price target to $155 in May, Matthew Ramsay of Canaccord Genuity wrote: “Our $155 price target is based on shares trading at the average of roughly 25x our base case [fiscal] 2021 ([calendar] 2020) non-GAAP EPS of $5.50 and roughly 25x our bull case estimate of $7.00.”

When this writer asked Ramsey why the price target is greater than the product of 25 and $5.50, he explained at the time that “We’ve treated our more conservative base case as more of a bearish outcome given the fundamentals I see currently” and acknowledged that “we should clean up the vocabulary.”

However, when Ramsay once again raised his price target in August, this time to $190 in a report entitled “The pullback folks have been waiting for,” he explained it thus: “Our $190 price target is based on shares trading at the average of roughly 25x our base case F2021 (C2020) non-GAAP EPS of $6.15 and roughly 27.5x our bull case estimate of $8.25.”

Of course, 25 times $6.15 is not $190, but $153.75 (before it is even discounted back to the present). When asked to clarify this time around, Ramsay did not immediately respond.

For a stock like Nvidia, one might assume that this type of numerical nitpicking is beside the point. That the stock is not rising on Friday because Muse floated a $10 earnings-per-share possibility, but because the analyst has a good sense of where the stock is going thanks to his industry expertise and exposure to management.

But Muse has not exactly had the hot hand on Nvidia. He had a hold rating starting at the end of 2014, when the stock was at about $20. He maintained that hold rating until December 2016, by which point the stock had quintupled to $100.

At this point, investors might worry that after underestimating the stock, he is now going a bit too far in the other direction.

On the other hand, they might be heartened by the potential that CEO Jensen Huang is indeed optimistic about his own company.