Goldman Sachs wants you to buy Nvidia. Again.
In a note circulated to clients this week, the investment bank reiterated its Conviction Buy rating on Nvidia and set a 12-month price target of $165, representing roughly 30% upside from recent trading levels near $130. The analyst behind the call, Toshiya Hari, argued that the chipmaker’s dominance in accelerated computing and artificial intelligence infrastructure remains unmatched, and that the current stock price doesn’t fully reflect the earnings trajectory ahead. As Yahoo Finance reported, Goldman’s message was blunt: the risk-reward on Nvidia is still skewed to the upside.
That’s a bold stance given the turbulence of the past several months.
Nvidia shares have been on a wild ride in 2025. After peaking above $149 in January, the stock cratered to roughly $86 in April amid a broader tech selloff fueled by tariff fears and questions about AI spending durability. It has since clawed back most of those losses, trading around $130 to $135 in recent sessions. The recovery has been sharp — but the scars remain. And Goldman is essentially telling clients that the pain was a buying opportunity, not a warning sign.
The core of Goldman’s thesis rests on Nvidia’s data center business, which has become the overwhelming driver of the company’s revenue. In the most recent quarter, Nvidia reported data center revenue of $26.3 billion, a staggering figure that nonetheless came in roughly in line with expectations. The Blackwell architecture — Nvidia’s latest generation of AI accelerators — is ramping aggressively, with management indicating that demand continues to outstrip supply. Goldman’s Hari projects that Nvidia’s earnings per share will reach $4.56 in fiscal year 2026 and climb to $5.82 in fiscal 2027, implying the stock trades at roughly 22 to 28 times forward earnings depending on which year you use. For a company growing revenue at triple-digit percentages, that multiple doesn’t look particularly stretched.
But not everyone on Wall Street shares Goldman’s conviction.
The bear case — or at least the cautious case — centers on a few interrelated concerns. First, there’s the question of whether hyperscaler capital expenditure can continue at its current pace. Microsoft, Amazon, Google, and Meta have collectively committed hundreds of billions of dollars to AI infrastructure buildouts. At some point, the argument goes, these companies will need to demonstrate returns on that spending. If they pull back, Nvidia’s revenue trajectory could flatten or even decline. Second, competition is intensifying. AMD’s MI300X chips have gained traction, and custom silicon efforts from Google (TPUs), Amazon (Trainium), and others are chipping away at Nvidia’s monopoly-like position. Third, there’s geopolitical risk. U.S. export restrictions on advanced chips to China have already cost Nvidia billions in potential revenue, and the regulatory environment remains unpredictable.
Goldman acknowledges these risks but argues they’re already reflected in the valuation. The bank’s note pointed to Nvidia’s software moat — specifically CUDA, the programming platform that has become the de facto standard for AI development — as a durable competitive advantage that makes switching costs prohibitively high for most customers. That’s a fair point. CUDA has been around for nearly two decades, and the developer community built around it numbers in the millions. Replicating that isn’t something AMD or any custom chip effort can do overnight.
Still, the CUDA argument has limits.
Google’s JAX framework and other open-source alternatives are gaining adoption, particularly among researchers and startups who aren’t locked into legacy Nvidia infrastructure. PyTorch, the most popular AI development framework, has been adding better support for non-Nvidia hardware. The moat is real, but it’s not impenetrable. And in a world where AI workloads are diversifying — from training massive models to running inference at the edge — the hardware requirements are shifting in ways that could open doors for competitors.
The tariff situation adds another layer of complexity. President Trump’s trade policies have whipsawed markets throughout 2025, and semiconductor stocks have been particularly sensitive. Nvidia generates significant revenue from China-adjacent markets, and any escalation in trade tensions could hit the stock hard. The April selloff was driven in part by fears that new tariffs would disrupt global supply chains and crimp demand for AI hardware. Those fears have eased somewhat, but they haven’t disappeared. According to recent reporting by Reuters, Nvidia disclosed earlier this year that U.S. export controls on its H20 chips to China would result in roughly $5.5 billion in charges — a material hit by any measure.
So why is Goldman still so bullish?
Part of the answer is simply math. Even if you haircut Nvidia’s growth projections significantly, the company is generating cash at an extraordinary rate. Free cash flow in the most recent quarter exceeded $15 billion. The balance sheet is fortress-like, with over $43 billion in cash and short-term investments against minimal debt. Nvidia has been returning capital to shareholders through buybacks — $27.7 billion authorized in the most recent program — which provides a floor under the stock even if growth decelerates.
And then there’s the Blackwell ramp. This is the part of the story that gets Goldman most excited. The Blackwell GPU architecture, which began shipping in volume in late 2024, represents a generational leap in performance for AI training and inference workloads. Early customer feedback has been overwhelmingly positive, and Nvidia CEO Jensen Huang has described demand as “insane” on multiple earnings calls. Goldman’s model assumes Blackwell will drive a significant step-up in average selling prices and margins over the next several quarters, which is the primary engine behind the firm’s earnings estimates.
There’s evidence to support that view. Microsoft recently expanded its Azure AI infrastructure with Nvidia’s latest chips. Meta has publicly committed to spending upward of $60 billion on AI infrastructure in 2025 alone, with Nvidia hardware forming the backbone of that investment. Amazon Web Services, despite its own custom chip ambitions, continues to offer Nvidia-based instances as its premium AI computing option. The hyperscalers are hedging — building their own chips while simultaneously buying everything Nvidia can produce. That dual-track approach suggests Nvidia’s TAM isn’t shrinking anytime soon.
But here’s where it gets tricky.
Nvidia’s stock isn’t cheap by historical standards, even if it looks reasonable on a growth-adjusted basis. The company’s market capitalization hovers around $3.2 trillion, making it one of the three most valuable companies on Earth alongside Apple and Microsoft. At that scale, the law of large numbers becomes a real constraint. To justify a $165 price target, Nvidia needs to keep growing at rates that would be extraordinary for any company, let alone one already generating over $100 billion in annualized revenue. Goldman’s implied upside assumes near-flawless execution on Blackwell, continued hyperscaler spending acceleration, and no major geopolitical disruptions. That’s a lot of things that need to go right.
Other analysts have been more measured. Bank of America maintains a Buy rating but with a lower price target. Morgan Stanley has flagged concerns about inventory build in the channel and the potential for a demand air pocket in the second half of 2025 as customers digest their Blackwell purchases. And some independent research firms have been outright bearish, arguing that the AI infrastructure buildout is entering a phase of diminishing returns where the next dollar of capex generates less incremental value than the last.
The broader market context matters too. The S&P 500 has recovered from its April lows, but sentiment remains fragile. Interest rates are still elevated relative to the post-2008 era. And the concentration of market gains in a handful of mega-cap tech names — Nvidia chief among them — has raised concerns about fragility. If institutional investors decide to rotate out of AI winners and into more defensive sectors, Nvidia could face selling pressure regardless of its fundamentals.
None of this means Goldman is wrong. The bank has been right on Nvidia for years, and its analysts have a strong track record in the semiconductor space. Hari’s note makes a compelling case that the market is underappreciating the duration and magnitude of the AI infrastructure cycle. If you believe — as Goldman clearly does — that we’re still in the early innings of a multi-year buildout that will reshape enterprise computing, telecommunications, autonomous vehicles, and scientific research, then Nvidia at 25 times forward earnings is a bargain.
That’s a big if.
The AI spending cycle has already lasted longer and run hotter than most analysts predicted two years ago. Jensen Huang’s vision of accelerated computing replacing general-purpose computing across the entire data center is ambitious and, if realized, would imply a total addressable market measured in the hundreds of billions of dollars annually. But technology cycles have a way of surprising even their biggest proponents. The dot-com era produced real companies and real technologies, but it also produced a speculative bubble that took years to unwind. The parallels aren’t exact — Nvidia’s profits are very real, unlike many dot-com darlings — but the pattern of extrapolating current growth rates indefinitely is a familiar trap.
For now, the smart money appears to be siding with Goldman, at least directionally. Nvidia remains the most widely held stock among hedge funds, according to recent 13F filings analyzed by Bloomberg. Short interest is minimal relative to the float. And retail investors continue to pile in, with Nvidia consistently ranking among the most-traded names on platforms like Robinhood and Schwab.
The question isn’t really whether Nvidia is a great company. It is. The question is whether it’s a great stock at this price, given the expectations already baked in. Goldman says yes. The market, with its $3 trillion-plus valuation, mostly agrees. But the margin for error is razor-thin, and the next few quarters of Blackwell revenue will either validate the bulls or give the skeptics their moment.
Growing up in the Midwest, I learned early that when everyone at the table agrees on something, it’s usually time to ask what you’re missing. Goldman’s call on Nvidia might be exactly right. But the unanimity of bullish sentiment around this stock — the sheer weight of conviction from banks, hedge funds, and retail traders alike — is itself a data point worth considering. The best trades are often the loneliest ones. Right now, being bullish on Nvidia is anything but lonely.
That doesn’t make it wrong. It just makes the stakes higher.
WebProNews is an iEntry Publication