Pandora (P) stock is hopping -- up 14% in just 17 days -- and you have Wall Street to thank for it.

In just eight days, Pandora is expected to report earnings for its fiscal second quarter, but Wall Street isn't waiting around for the good news. Three weeks ago we saw Morgan Stanley's upgrade of Pandora stock, based on an optimistic belief that the arrival of on-demand subscription radio will deliver beaucoup bucks to the internet radio star.

This morning, two new analysts are tuning into the same story -- and they like what they hear.

Here's what you need to know.

Wall Street is grooving to Pandora's plan for on-demand radio. Image source: Getty Images.

1. Needham raises Pandora from like to love

This morning, analysts at Needham jumped on the Pandora soul train, raising their price target on the already buy-rated stock by one-third, to $16 a share. As explains, Needham has just updated its estimates for Pandora's fiscal 2017 performance, and thinks the prospects look good.

Mind you, Needham still expects to see Pandora lose money (and rather a lot of it) this year. But the analyst also believes that next year's news will feature "new revenue ... from on-demand services, margin expansion upside," and even "a potential sale" of the business to an acquirer. Those prospects have Needham firmly on the buy side of this stock -- and Needham's not alone.

2. Piper loves it, too

Needham's hike in price target, while good news, isn't quite as exciting as an actual upgrade of the stock, of course. But Pandora has that, too. No sooner had Needham come out with its endorsement, you see, than Piper Jaffray announced that it has decided to upgrade Pandora stock to overweight -- and Piper assigned a higher price target than either Needham (or Morgan Stanley before it) had ventured to propose: $18 a share.

According to Piper, Pandora is all but certain to release an on-demand music subscription service soon, and when it does, this will add 9.2 million new paying customers to the service (which today, largely subsists on non-paying users consuming ads). Piper further believes that these customers will generate "$750M+ in revenue and $120M in EBITDA over the next 3 years."

3. Is that realistic?

On the surface at least, the numbers do seem plausible. $750 million in revenue from 9.2 million paying customers over three years? That's only $27 a head annually -- hardly an unreasonable revenue goal from a fee-for service, even accounting for customer churn, and the likelihood that the service would begin with far fewer than 9.2 million customers, and only gradually build to Piper's hoped-for levels of adoption.

For that matter, 9.2 million customers itself doesn't seem an unreasonable goal. That's only 11.6% of the 79.4 million "active listeners" that Pandora claimed at the end of Q1. Offer enough free trials, and enough "Act now to claim this discount offer!" advertisements, and I could see how Pandora might convince about one in 10 members of its installed base to fork over a few dollars for on-demand music.

The most important thing: Valuation

But all of this still doesn't address the most important question: Even assuming Piper Jaffray, Needham, and Morgan Stanley are 100% right in their predictions about how the business will play out -- and even assuming Spotify, YouTube, and all its other rivals play dead and do nothing to disrupt Pandora's grand plan -- will that be enough to make Pandora stock a buy?

I don't think so.

Here's why: According to Piper Jaffray, the introduction of on-demand music subscriptions will be enough to almost single-handedly lift Pandora stock out of a sea of red ink, and safely into the black. The analyst posits $0.02 per share in earnings next year, and $0.36 per share in profits in fiscal 2018.

That's an amazing growth rate projection of 1,800% in just one year (and all the more amazing because most analysts expect no more than 26% annualized earnings growth out of Pandora over the next five years). If Piper is right, then a stock that has no P/E today (because it has no earnings), and costs 650 times next year's hoped-for profits, will drop to a P/E of just 36 just two years out. It's a buy thesis that seems too good to resist.

But it leaves investors with just one big question.

Pandora has lost money every year it's been in business. According to data from S&P Global Market Intelligence, it lost $170 million last year, lost $236 million over the past 12 months, and is burning cash at the rate of $124 million per year.

So...if Pandora really can avoid all those losses, and become instantly, presto-change-o profitable just by adding a subscription model next year...why hasn't it done so already?