AMD at $514, Palantir at $174: Which Price Holds Up

24/7 Wall St. weighed AMD at $514 and Palantir at $174 on Aug. 17, 2026, arguing one of the two AI bellwethers sits in a constructive setup into 2027 while the other's valuation already prices in two years of flawless execution; AMD last traded at $515.24 (+0.17%) and Palantir at $175.19 (+0.66%) at 15:24 GMT.

The AI trade has stopped being a single position. That is the useful takeaway from a Monday comparison of two of its most-owned names: AMD (AMD) at roughly $514 and Palantir (PLTR) at roughly $174, price levels that look superficially similar in their momentum and behave very differently once you ask what each one requires from the next eight quarters.

By mid-session on Aug. 17, 2026, both were higher. AMD last changed hands at $515.24, up 0.17% from a prior close of $514.39, having traveled between $506.33 and $517.35 on the day. Palantir was at $175.19, up 0.66% from $174.04, with an intraday band of $172.30 to $176.28. Those are live prices as of 15:24 GMT, and they are the same neighborhood the headline figures come from — this is a story about valuation, not about a move.

Two AI anchors, one uncomfortable question

Both companies have earned the label of AI infrastructure proxy, and both trade like it. But the framing put forward by 24/7 Wall St. is that only one of the two is in what it calls a constructive setup heading into 2027, while the other carries a multiple that has already discounted two years of flawless execution. That distinction matters more than the direction of AI spending, because the direction of AI spending is no longer the contested part of the thesis.

Think of it as two separate bets stacked on top of the same macro view. The first bet is that hyperscalers, sovereign programs and enterprise buyers keep committing capital to compute and to software that sits on top of it. The second bet is that the price you paid leaves room for the company to merely do well rather than perfectly. A stock can win the first bet and still lose money for its holders on the second.

What a "constructive setup" actually has to mean

For a semiconductor business, a constructive setup is mechanical rather than narrative. It means visible capacity commitments, a product cadence that lands on schedule, gross margin that improves as the mix shifts toward accelerators, and a customer list broad enough that the loss of any single design win does not reset the model. Chip cycles are lumpy; the reason investors will pay for one is that the revenue arrives in identifiable chunks tied to shipments they can count.

For an enterprise software business, the equivalent test is different. Growth has to come from expanding contracts inside existing accounts as much as from new logos, and the operating leverage has to be real — revenue growing faster than headcount and sales cost, quarter after quarter, not once. Software multiples survive on the belief that the next dollar of revenue is nearly free. When that belief cracks, the de-rating is fast, because the multiple was the asset.

Neither company is short of demand signals in the current cycle. The argument being made is narrower: that at $514 one of these two still has a plausible gap between what the market assumes and what the business can deliver, and that at $174 the other has closed it.

Reading the tape around them

Monday's index prints put the individual moves in context. The Nasdaq 100 tracker QQQ was at $734.32, up 0.44%, while the S&P 500 tracker SPY sat at $775.76, down 0.07%, and the Dow tracker DIA was at $535.11, down 0.31%. That is a session with a mild growth tilt and nothing else — money leaning toward large-cap tech while the broader and more industrial parts of the market gave back ground.

Both AMD and Palantir traded in bands of roughly the same relative width. On illustrative arithmetic from the quoted ranges, AMD's $506.33-to-$517.35 spread works out to about $11.02, or roughly 2.1% of its prior close; Palantir's $172.30-to-$176.28 spread is about $3.98, or roughly 2.3%. In other words, on the day the market was not treating one as materially riskier than the other. The divergence the comparison points to is not something you can see in a single session's volatility — it is embedded in the multiple, and multiples only show up in the tape when something forces a revision.

Where the risk actually sits

In other words, on the day the market was not treating one as materially riskier than the other.

The uncomfortable feature of a valuation that assumes two years of perfect execution is that it removes the ordinary cushion. A quarter that beats but guides conservatively, a hiring plan that runs ahead of bookings, a large government contract that slips a quarter for procurement reasons — none of that is a broken business, but all of it is enough to compress a premium multiple. Investors who bought the AI story rather than the price are the ones who tend to discover this after the fact.

On the semiconductor side the risks are more familiar and, arguably, easier to underwrite: supply allocation, packaging capacity, the cost of competing for the same advanced nodes, and the possibility that customers designing their own silicon eventually claw back volume. Those risks are cyclical and countable. They also tend to be at least partly reflected in the price, which is exactly what "constructive setup" is meant to convey.

What to watch into 2027

Three things will settle this debate rather than any single price target.

Accelerator revenue disclosure. Whether AI-specific chip revenue keeps being broken out clearly, and whether the sequential slope steepens or flattens, is the cleanest read on whether the hardware bet is on schedule.
Net revenue retention and contract duration in software. Expanding existing accounts at rising dollar values is what justifies a premium multiple; new-customer counts alone do not.
Operating margin direction. For both names, the market is paying for leverage. A stalled margin line is the earliest visible sign that the flawless-execution assumption is slipping.

None of that requires a call on whether AI capital spending continues. It almost certainly does. The question posed by these two prices is whether the spending accrues to shareholders at the level they paid, and on that, $514 and $174 are not equivalent starting points. Investors who own both should at least know which of the two positions is the valuation risk in the sleeve — and be honest that holding a name because the underlying trend is real is not the same as holding it because the price is reasonable.

Key facts

AMD last trade: $515.24, +0.17% (as of 15:24 GMT, Aug. 17, 2026)
PLTR last trade: $175.19, +0.66% (as of 15:24 GMT, Aug. 17, 2026)
Prior closes: AMD $514.39; PLTR $174.04
Index backdrop: QQQ $734.32 (+0.44%), SPY $775.76 (-0.07%), DIA $535.11 (-0.31%)

Frequently asked questions

What prices are AMD and Palantir trading at?

As of the last trade at 15:24 GMT on Aug. 17, 2026, AMD was at $515.24, up 0.17% from a prior close of $514.39. Palantir was at $175.19, up 0.66% from a prior close of $174.04. AMD's intraday range was $506.33 to $517.35; Palantir's was $172.30 to $176.28.

What is the core argument in the AMD versus Palantir comparison?

24/7 Wall St. framed the two as both anchoring the AI trade, but with one sitting in a constructive setup heading into 2027 and the other carrying a valuation that has already priced in two years of flawless execution. The distinction is about valuation cushion, not about whether AI demand continues.

Why does a high multiple matter if AI demand is still strong?

Because the price sets the bar. A premium multiple assumes near-perfect delivery, so an in-line quarter, cautious guidance or a contract slipping by a quarter can compress the valuation even when the underlying business is healthy. Investors can be right about the trend and still lose money on the entry price.

How did the wider market trade that day?

The session had a mild growth tilt. The Nasdaq 100 tracker QQQ rose 0.44% to $734.32, while the S&P 500 tracker SPY slipped 0.07% to $775.76 and the Dow tracker DIA fell 0.31% to $535.11. Both AMD and Palantir were modestly higher against that backdrop.

What metrics would confirm the semiconductor side of the thesis?

Clear disclosure of AI accelerator revenue and the direction of its sequential slope, gross margin improving as the product mix shifts toward accelerators, a product cadence that lands on schedule, and a customer base broad enough that losing any single design win does not reset the model.

What should software investors watch instead?

Expansion inside existing accounts — rising dollar value per customer — rather than new-logo counts alone, plus contract duration and the direction of operating margin. Premium software multiples rest on the belief that the next dollar of revenue is nearly free, so a stalled margin line is the earliest warning sign.

Sources

AMD at $514, PLTR at $174: Buy, Sell or Hold? — 24/7 Wall St