Why Three Leading Wall Street Analysts Remain Bullish on Nvidia Shares

Chipmaker Nvidia (NVDA) is widely viewed as one of the biggest winners of the artificial intelligence revolution, supported by strong and sustained demand for its high-performance graphics processing units. These GPUs remain essential for training and running advanced AI models, placing Nvidia at the center of the industry’s rapid expansion.

‍

However, Nvidia’s stock has faced pressure in recent weeks as investors reassess valuations across AI-related names. Concerns have also grown around intensifying competition in the AI chip market, with rivals such as Broadcom (AVGO), Advanced Micro Devices (AMD), and Alphabet-owned Google promoting their own custom silicon, including tensor processing units.

‍

Adding another layer of uncertainty, Nvidia continues to navigate restrictions and regulatory risks tied to exporting advanced chips to China amid ongoing geopolitical tensions between Washington and Beijing.

‍

Even with these headwinds, several leading Wall Street analysts remain constructive on Nvidia’s long-term outlook. Their optimism is rooted in the company’s proven execution, deep technology moat, relentless pace of innovation, and commanding share of the AI GPU market. TipRanks’ AI Analyst also maintains an “outperform” rating on NVDA, assigning a price target of $205.

‍

Below is a closer look at three well-regarded analysts who continue to see meaningful upside in Nvidia shares.

‍

Vivek Arya – Bank of America

After hosting a virtual meeting with Nvidia’s vice president of investor relations, Toshiya Hari, Bank of America analyst Vivek Arya reiterated his buy rating on NVDA and reaffirmed a price target of $275. He continues to rank Nvidia among his top investment ideas.

‍

Arya noted that while Nvidia acknowledges Google’s Gemini 3 as a leading large language model trained on Google’s in-house TPUs, management believes it is far too early to crown a long-term winner.

‍

Importantly, Nvidia pointed out that current GPU-based LLMs were trained using its older Hopper architecture from 2022. As a result, they should not be compared directly with next-generation models expected to be trained on Nvidia’s Blackwell GPUs, introduced in 2024.

‍

According to Arya, Nvidia’s leadership is confident that LLMs powered by Blackwell will begin rolling out in early 2026, demonstrating that the company remains at least a full technology cycle ahead of competitors.

‍

Independent benchmarks such as MLPerf and InferenceMAX already support this view, ranking Blackwell as the top performer in both training and inference workloads. Nvidia also leads on efficiency metrics like tokens per watt and revenue per token.

‍

The five-star analyst added that Nvidia sees demand and supply visibility for at least $500 billion in cumulative revenue opportunities tied to Blackwell, Rubin, and networking products across calendar years 2025 and 2026.

‍

Notably, recent agreements with OpenAI and Anthropic/Microsoft are incremental to this forecast, as they were structured as letters of intent and could represent additional upside.

‍

Arya concluded that the discussion reinforced his bullish stance, particularly given Nvidia’s valuation. Based on projected earnings, NVDA trades at roughly 25 times 2026 earnings and 19 times 2027 earnings, implying a PEG ratio of just 0.5x. That compares favorably with an average PEG closer to 2x for the Magnificent Seven and other high-growth peers.

‍

Stacy Rasgon – Bernstein

Bernstein analyst Stacy Rasgon also remains positive on Nvidia, maintaining a buy rating and a $275 price target. In his latest investor note, Rasgon shared insights from a virtual meeting with Nvidia’s senior director of investor relations, Stewart Stecker.

‍

Rasgon highlighted that Nvidia’s previously announced $500 billion revenue outlook for cumulative Blackwell, Rubin, and networking sales in 2025 and 2026 could ultimately prove conservative. The forecast excludes newer arrangements, including Nvidia’s partnership with Anthropic, its 10-gigawatt collaboration with OpenAI, and several initiatives in the Middle East.

‍

Addressing competition from Google’s custom chips, Rasgon said Nvidia recognizes Google’s progress over more than a decade but believes it remains roughly two years ahead of the TPU program. Nvidia argues that as AI models evolve rapidly, it will be difficult for Google to convince cloud providers to broadly deploy TPUs, which are optimized for specific architectures. By contrast, Nvidia’s programmable platforms are seen as more flexible and better suited for cloud-scale AI infrastructure.

‍

Rasgon also commented on President Donald Trump’s recent remarks suggesting Nvidia could export H200 AI chips to China if 25% of related revenue is shared with the US government. He noted that Nvidia is still awaiting export licenses for the H200, after which it would assess demand and begin production. The company has yet to receive formal guidance on how the proposed 25% revenue share would be implemented or accounted for.

‍

Blayne Curtis – Jefferies

Jefferies analyst Blayne Curtis reiterated a buy rating on Nvidia with a price target of $250 in his outlook for the semiconductor sector in 2026. While Curtis named Broadcom as his top overall pick due to expected gains from ASIC adoption, he emphasized that he remains firmly bullish on Nvidia.

‍

Curtis argued that concerns around Nvidia are overstated, pointing to the company’s durable technology advantage and valuation of roughly 18 times a $10 earnings-per-share benchmark. He believes ASIC adoption is still in its early stages, leaving Nvidia ample room to benefit from continued AI infrastructure spending.

‍

He also noted that the Blackwell Ultra rollout remains on schedule, with the Rubin platform expected to ramp in the second half of 2026. Additional catalysts include the launches of Vera-Rubin and NVLink 6 in late 2026, which Curtis expects will further strengthen Nvidia’s competitive position. Blackwell-based LLMs are anticipated in the first half of 2026, serving as a potential stock catalyst.

‍

Curtis also sees Nvidia’s upcoming CPX chip as another growth driver, particularly as hyperscalers increase capital spending and place greater emphasis on inference. He projects CPX revenue of $13 billion in 2027 and raised his EPS estimates for Nvidia to $7.82 for 2026 and $9.50 for 2027.