• Bernstein reaffirms its Outperform rating and $275 price target on NVIDIA (NVDA), emphasizing the company's structural lead in AI data-center chips.
  • Near-term risks include U.S. export controls delaying H200 shipments to China and uncertainty over a proposed 25% government revenue share on some sales.
  • NVIDIA maintains it is roughly two years ahead of Google (GOOG)'s TPU efforts, with its programmable GPU platform better suited for evolving AI models and cloud workloads.

Bernstein has doubled down on its bullish stance for NVIDIA, reiterating an Outperform rating and a $275 price target in a recent analysis. The firm argues that despite looming regulatory hurdles, NVIDIA's technological edge in AI accelerators remains formidable, driven by a deep software ecosystem and rapid innovation cycles that competitors struggle to match.

According to people familiar with the matter, NVIDIA is still awaiting U.S. approval to ship its high-end H200 chips to China, a key market that has historically contributed significantly to its data-center revenue. This delay comes amid broader U.S.-China tech tensions, with export controls tightening over advanced semiconductors. Bernstein notes that without clarity on these approvals, NVIDIA's near-term growth in China could face headwinds, though the company has previously navigated such restrictions by redesigning products like the A800 and H800 variants.

In addition to export challenges, there is uncertainty over a proposed 25% government revenue share tied to certain China-related sales, which could impact profitability if implemented. NVIDIA has not publicly commented on this potential policy, but sources indicate that discussions are ongoing with regulatory bodies. Attempts to reach NVIDIA for further comment were unsuccessful at press time.

On the competitive front, Bernstein highlights NVIDIA's view that it remains about two years ahead of Google's TPU efforts, citing the flexibility of its programmable GPU platform as a critical advantage for fast-evolving AI models and cloud infrastructure. "What institutional investors are really focused on is technological stability and ecosystem depth," one analyst paraphrased, pointing to NVIDIA's CUDA software and partnerships with hyperscalers as barriers to entry. While acknowledging progress from competitors like TPUs, the firm believes NVIDIA's hardware is better suited for the dynamic demands of AI workloads, with cloud providers continuing to standardize on its platforms.

Financially, NVIDIA recently reported quarterly revenue growth exceeding 60% year-over-year, driven by surging data-center AI demand, with net margins above 50% and return on equity near 100%. Wall Street consensus remains bullish, with an average 12-month price target in the mid-$250s, though Bernstein's $275 target implies further upside. Market data shows NVIDIA's stock trading with volatility amid these regulatory uncertainties, but analysts see the AI data-center build-out as still in an early phase, with trillions in infrastructure spending projected by 2030.

Looking ahead, short-term factors hinge on U.S. approvals for H200 exports and the final shape of revenue-share policies. Long-term, Bernstein forecasts continued upside as AI workloads proliferate, calling the data-center opportunity "enormous" and emphasizing NVIDIA's sustainable model. Other firms, including Goldman Sachs (GS), have recently reaffirmed similar bullish ratings, reinforcing NVIDIA's position as a core AI infrastructure play despite geopolitical pressures.

Correction: An earlier version of this article misstated the specific chip models affected by export controls; it has been updated to clarify that H200 shipments are pending approval, while previous variants like A800 and H800 were redesigned for compliance.