“Cheap” is the slipperiest word in investing. The word only describes a price sitting next to somebody’s forecast, and the forecast does most of the work.
That is why the label keeps landing on stocks that have already run hard. A company doubles its revenue, the share price climbs to match, and the multiple ends up close to where it started.
Nothing got less expensive in the way a shopper would recognize. The denominator simply ran faster than the numerator.
The artificial intelligence (AI) trade has lived inside that logic for two years. Chip stocks absorbed enormous gains while analysts kept telling clients the shares were reasonable, because profits were climbing faster than prices. That argument has been right often enough that doubting it has cost people real money.
So the question worth asking is not whether the biggest chip stock has gotten expensive. It is what happens when a firm applies the word cheap to a company already worth more than $5 trillion, then attaches a price target most of Wall Street would call conservative.
That combination landed this week. Piper Sandler started coverage of Nvidia (NVDA) with an overweight rating and a $300 price target, calling the company the “outright leader in AI compute,” according to Investing.com.
Piper Sandler says $5 trillion Nvidia is still underpriced, sets Nvidia target 8% below consensus.SOPA Images / Getty Images
How Nvidia’s earnings outran its own share price
Before any of the valuation math makes sense, the scale of what Nvidia just reported has to register.
Revenue for the quarter that ended July 26 came in at $96.2 billion, up 106% from a year earlier, according to Nvidia. Data center revenue accounted for $89.0 billion of that, and gross margin held at 75%.
More Nvidia:
Adjusted earnings landed at $2.22 a share for the quarter. For all of fiscal 2026, which closed in January, adjusted earnings were $4.77 a share.
Set those two figures side by side and the shape of the story appears. A single quarter now delivers close to half of what an entire year produced eight months earlier.
The company returned roughly $26 billion to shareholders during the quarter through buybacks and dividends, and still had about $99 billion left on its repurchase authorization.
That is the background that lets an analyst describe a $5 trillion company as inexpensive without laughing. The share price has roughly kept pace with the profits. It has not obviously outrun them.
What Piper Sandler’s new Nvidia rating actually says
Analyst David O’Connor started the stock at overweight as part of a broader initiation across AI chipmakers, and the specific numbers behind the call matter more than the rating.
Nvidia holds roughly 80% of the AI compute market and about 50% of unit shipments, the firm estimated, according to Investing.com. Supply across the industry should stay constrained for another two to three years, the note said, with agentic AI workloads driving a step change in demand.
From there the model gets aggressive. Piper Sandler projects Nvidia revenue growing at a 47% compound annual rate through fiscal 2030, with earnings reaching $30 a share.
I ran that estimate against the company’s own reported results. Getting from $4.77 in adjusted earnings for fiscal 2026 to $30 by fiscal 2030 requires profit compounding at roughly 58% a year for four straight years, with no stumble.
The $300 target comes from applying a 14 times multiple to calendar 2028 estimates. At that multiple the analyst placed Nvidia among the least expensive names in the AI complex, according to Investing.com.
The broader sector call rested on what the firm described as “insatiable demand” for compute capacity, according to CNBC.
Strip away the adjectives and the call reduces to a single bet. Pay $300 today and you are paying about 10 times the earnings Piper Sandler expects to arrive in 2030.
Where the $300 Nvidia target sits against Wall Street
Most of the coverage around this call has skipped the comparison that matters. A $300 target sits below the crowd rather than above it.
Piper Sandler set its target at $300, roughly 34% above the prior close, according to CNBC.
BMO Capital carries a $340 target and an outperform rating, built on a 70% growth projection for fiscal 2028, according to Investing.com.
The average 12-month target across analysts covering the stock stands at $327.18, with 58 buy ratings against one sell, according to Investing.com’s Nvidia quote page.
Wedbush’s Dan Ives set a $250 base case for the end of 2026, according to TheStreet.
JPMorgan reset its Nvidia target to $265 for the rest of 2026, according to TheStreet.
What struck me in my analysis of that target board is the gap between the language and the arithmetic. Piper Sandler used the most bullish vocabulary available about Nvidia’s competitive position, then published a number about 8% below what the average analyst already expects.
The firm is bullish on the business and disciplined about the multiple, which is a different posture than the one most Nvidia notes take.
Other banks have justified higher targets partly by stretching the multiple they are willing to pay. Piper Sandler arrived at a lower number by holding that multiple down.
For anyone holding the stock, that distinction changes what the call is worth. A target resting on a modest multiple has room to survive a market that decides to pay less for growth, which is exactly the scenario that has punished AI names during every drawdown since 2024.
What the Groq antitrust probe could do to the math
The initiation landed the same week as a problem that no earnings model captures.
The Justice Department is investigating whether Nvidia structured its roughly $20 billion licensing agreement with AI chip startup Groq to avoid antitrust review, according to Bloomberg. The arrangement licensed Groq’s inference chip technology and brought over several of its executives, including founder Jonathan Ross, without a formal acquisition.
Regulators sent a formal demand for information and the inquiry remains open, according to Bloomberg. Nvidia shares slipped about 1% on the report.
The near-term financial risk looks limited, since officials do not currently expect the deal to be unwound. The longer-term question is structural. Nvidia has used licensing, hiring and minority investment to absorb capability across the AI stack, and a ruling that this pattern requires merger review would slow the pace of that expansion.
Piper Sandler’s 47% growth rate assumes Nvidia keeps converting demand into revenue at the speed it has managed for three years. Regulatory friction is one of the few things that can interrupt that conversion without any change in demand.
Nvidia guided to $108 billion in revenue for the current quarter, marking another sequential step of nearly $12 billion. That report, due in November, is the next real test of whether a 14 times multiple on 2028 earnings was conservative or optimistic.
Until then, investors have an unusual situation to sit with. My read is that the most valuable company on earth now has a Wall Street bull arguing it is underpriced, and that bull is asking for less than the consensus already expects.
