Just caught something interesting about two major AI growth stocks that Wall Street is getting pretty cautious on right now. Both Palantir and Micron have had incredible runs — we're talking nearly doubled and quadrupled respectively over the past year — but some serious analysts are pumping the brakes hard.

Let me break down what's happening here. Palantir's the one that caught my attention first. The company's fundamentals actually look solid on paper. Their AI and analytics software has real differentiation through this ontology-based architecture thing that gets smarter as it processes more data. Forrester and IDC both recognize them as leaders in their space. Q4 numbers were genuinely impressive too — 70% revenue growth hitting $1.4 billion, and they posted a Rule of 40 score of 127%, which basically never happens in software.

But here's where it gets wild. Brent Thill over at Jefferies just set a price target of $70 per share. If you do the math on that against the current $157 price, we're looking at potential 55% downside. And that's from a serious analyst, not some random on Twitter. The valuation is the killer — trading at 209 times adjusted earnings. Yeah, earnings are expected to grow 57% annually through 2027, but even for an AI growth stock with a real competitive moat, that's getting pretty expensive.

Then there's Micron. Memory chips are absolutely critical for AI infrastructure, and Micron's positioned as the third-largest DRAM and NAND supplier. They've been gaining share while Samsung lost ground. First quarter was brutal in the best way — revenue up 56% to $13.6 billion, earnings up 167%. William Kerwin at Morningstar has them at a $225 target, implying 40% downside from the current $380 level.

Here's the thing though — and this is crucial for anyone looking at AI growth stocks right now — Micron's gains are riding on a memory chip supply shortage. That's temporary by definition. The industry is cyclical, and when that shortage flips to oversupply, prices will crater. Kerwin points out that memory chips are basically commoditized, meaning there's no real economic moat protecting margins. Wall Street's own forecasts show earnings accelerating through fiscal 2027, then dropping sharply through 2029. That's the market already pricing in the cycle peak.

So what's the takeaway? Palantir's got genuine competitive advantages and real growth, but at current valuations one bad quarter could trigger a sharp correction. Micron's riding a temporary supply cycle that won't last. Neither stock is necessarily a bad business, but the risk-reward doesn't look great at these prices for either of them. If you're holding these, maybe keep your positions lean. Both could definitely surprise on the upside, but the downside scenarios analysts are flagging are worth taking seriously.