Nvidia Beats, Stock Dumps—BofA Says Buy the Dip

Of course, agentic AI is the miracle everyone on Wall Street is talking about right now.

Why BofA is still bullish

The core argument isn't about one quarter. It's about the size of the market Nvidia is selling into, and how fast that market is growing.

BofA previously estimated the total AI market at $1.7 trillion. They now expect it to grow 4x, reaching over $3 trillion by 2030. Within that, they model Nvidia holding roughly 78% of the AI accelerator market—the chips specifically built to run AI workloads. That's basically a near-monopoly in the fastest-growing technology market BofA says it has ever tracked.

There's also a newer opportunity the bank recently upgraded. Nvidia is moving into agentic CPU chips: processors designed for AI agents, the kind of software that can autonomously complete complex tasks without human input. BofA raised its estimate for that market from $125 billion to $200 billion, and says Nvidia already has $20 billion in demand locked in for the second half of this fiscal year.

The demand isn't speculative, either. Customer purchase commitments totaled $145 billion this quarter, up from $95 billion just three months ago. AWS alone has committed to deploying around 1 million Nvidia GPUs through 2027. These are contracts, not wishlists.

The risk

BofA lists six formal risks. Two deserve attention from anyone holding or considering Nvidia stock.

The first is the stock's own gravitational pull. Nvidia now represents 8.3% of the entire S&P 500 index—the benchmark that tracks America's 500 largest public companies. About 78% of active fund managers already own it. When that many people already hold a stock, there's simply a smaller pool of potential new buyers to push the price higher.

The second is custom chips. Big hyperscale cloud companies like Google, which recently launched its eighth-generation AI chips specifically designed to reduce dependence on Nvidia, are investing heavily in alternatives built in-house. BofA's counter: It still expects Nvidia to hold more than 70% of the accelerator market long-term, arguing that full-platform support and AI factory infrastructure are things custom chips simply can't replicate.

There's also a recurring criticism that Nvidia's investments in companies like OpenAI and Anthropic—as a vendor selling them chips—amount to circular spending.

What the numbers look like from here

BofA raised its earnings-per-share estimates by 9% for fiscal 2027 (to $9.09) and 15% for fiscal 2028 (to $13.27). Put another way: Nvidia earned $4.55 per share last fiscal year. BofA expects that to roughly double to $9.09 this year, then hit $13.27 the year after. Earnings per share growing at 43% annually is rare for any company, let alone one already worth $5.5 trillion.

At its current price, Nvidia trades at 19.7 times its estimated 2027 earnings. A metric that adjusts the P/E for growth rate, where lower is better—sits at 0.5x against a Mag-7 average of 3.9x.

Free cash flow is projected to go from $96.7 billion last fiscal year to $186.8 billion in 2027 and $282 billion in 2028. The company also raised its quarterly dividend 25-fold, from $0.01 to $0.25 per share, and announced an $80 billion additional share buyback authorization—bringing total repurchase capacity to roughly $120 billion.

The $350 price target is based on 26x estimated 2027 earnings, within Nvidia's historical range of 25x to 56x. The next concrete date on the calendar: CEO Jensen Huang's keynote at Computex on June 1, where BofA expects him to lay out Nvidia's agentic AI roadmap and CPU strategy in detail.