NVIDIA (NVDA) Price on Solana

UBS reiterated its buy rating on Nvidia on Thursday, naming the stock a top idea heading into the fourth quarter. The designation signals heightened conviction from the bank as Q4 begins, a period that historically sees increased enterprise and cloud capex commitments.

The call adds to a string of constructive Wall Street views on Nvidia in recent weeks. UBS's top-idea label typically implies the firm expects the stock to outperform peers over the near term, reinforcing its position as a preferred name among large-cap AI infrastructure plays.

Nvidia's board approved a $150 billion increase to the company's share repurchase program on September 28, bringing the total remaining authorization to $235 billion. Nvidia calls it the largest share repurchase authorization increase in history and expects to complete the remaining program through fiscal 2028. CEO Jensen Huang said the company's "cash generation gives us the capacity to invest in the technologies that advance this transformation and return capital to shareholders," tying the move to what he described as a once-in-a-generation shift to AI and accelerated computing.

The expansion comes a month after Nvidia reported $96.2 billion in second-quarter revenue, up 106% year over year, according to The Hill. Nvidia shares rose about 2.3% in Monday morning trading after the announcement, and the stock is up roughly 24% this year. The larger authorization gives Nvidia room to keep returning cash from its AI data center business to shareholders while it goes on funding new chip and infrastructure development.

Nvidia on September 28 launched its Open Agent Safety Platform, a combined software and hardware stack for setting limits on what AI agents can do and stopping them when they cross those limits. The platform has two main parts. OpenShell is an open-source runtime that puts a secure boundary around agents running on CPUs. It is built for Nvidia's Vera CPU and can be extended to Arm and Intel platforms. Sentry is a separate watchdog that runs on Nvidia's BlueField-4 DPUs. It monitors how agents behave and, according to Nvidia, can quarantine one that breaks policy within milliseconds, using enforcement built into the silicon.

The launch ties Nvidia's AI-safety pitch directly to its own chips. Sentry runs on BlueField-4, and OpenShell is designed around Vera. Nvidia says more than 100 organizations back the platform, including Anthropic, Microsoft, Cisco, CrowdStrike, Palo Alto Networks, JPMorganChase, Salesforce, SAP, ServiceNow and Hugging Face. CEO Jensen Huang said AI's potential "will only be realized if we solve AI safety." The software is available through Nvidia's developer resources and GitHub.

Bank of America has put Nvidia on a list of buy-rated stocks it considers undervalued, telling investors to buy any weakness in the shares, CNBC reported on September 26. The bank keeps a $350 price target on Nvidia. That target is based on 22 times projected 2027 earnings, and the bank points to Nvidia's position in AI infrastructure as the main support for its view. Other names on the list include Dutch Bros, Thor Industries, Natural Grocers and Lyntris.

The $350 target is not new. BofA analyst Vivek Arya reiterated a Buy rating at that level in July, arguing that the market's worries about margin pressure, custom-chip competition and capital allocation were overdone. This latest call sits against a split week of Wall Street views on Nvidia: Bernstein recently set a $400 target, while Seaport reiterated its Sell rating.

Seaport reiterated its sell rating on Nvidia on September 25, according to CNBC's roundup of Friday's biggest analyst calls. The roundup did not include a price target. The firm argued Nvidia has "no easy way to surprise to the upside," because its output is limited by TSMC manufacturing capacity and by the law of large numbers at its current scale.

Seaport also flagged growing competition. It said the pressure has reached the point where Nvidia is "leaning heavily on its balance sheet and a bundling of its products to drive numbers."

Xi Jinping arrived at the White House for his first U.S. state visit in more than 11 years with a run of domestic AI chip launches behind him, but U.S. export controls on chips, the rules that keep Nvidia's most advanced AI processors out of China, are not on the summit agenda, according to U.S. Trade Representative Jamieson Greer. CNBC reports that Huawei pulled its next AI chip forward by three quarters, Alibaba unveiled what it called the country's most powerful AI processor, and DeepSeek reportedly assembled the largest known Huawei chip cluster to date. Lizzi Lee of the Asia Society Policy Institute put Beijing's message this way: the leverage is not "we have caught up with you" but "you can't simply technologically choke us off."

For Nvidia, the gap is still large. Each Chinese chip delivers roughly half the compute of Nvidia's, Huawei's SuperPoD design links fewer processors than planned, and Huawei chairman Eric Xu has acknowledged possible supply shortages. Chris Miller said American firms earn 50 to 100 times the revenue of comparable Chinese rivals. Still, with controls left off the table, the restrictions keep limiting Nvidia's access to China while domestic alternatives improve. Sigrid Wang said the U.S. rules "slow China's technological development but have not prevented" alternatives from emerging.

Bernstein analyst Stacy Rasgon reiterated an Outperform rating on Nvidia with a $400 price target — implying roughly 75% upside — in a note to clients published Monday. Rasgon's bull case rests primarily on valuation: Nvidia currently trades at approximately 17 times next-twelve-months earnings per share estimates, less than half its five-year average of around 35 times. He told clients he would "still own both Nvidia and Broadcom," adding that both stocks are "extremely cheap" and "appear to have line of sight for strong growth even in a constrained environment." The data center opportunity, in Bernstein's view, "remains enormous."

The call is more aggressive on valuation grounds than Wall Street consensus but directionally aligned with the prevailing buy-side sentiment that Nvidia's AI infrastructure dominance justifies a premium multiple. Where Rasgon diverges is in framing the stock as undervalued on near-term earnings rather than a growth-at-any-price trade — an argument that becomes more persuasive if hyperscaler capex holds up through 2027 Vera Rubin deployments.

Alibaba unveiled the Zhenwu V900, its latest in-house AI accelerator, at the company's annual Apsara Conference on September 22, claiming the chip delivers three times the performance of its predecessor, the Zhenwu M890, which entered market in May. The V900 is scheduled for mass production and commercial release in Q1 2027. Alongside the chip reveal, Alibaba Cloud outlined plans to scale global data center capacity to over 20 gigawatts by 2032, backed in part by an HK$80 billion equity placement completed in August — roughly 60 percent of which is earmarked for computing infrastructure expansion and hyperscale AI data center construction.

For Nvidia, Alibaba's custom silicon push is a meaningful competitive signal from one of the world's largest AI accelerator buyers. A proprietary chip targeting enterprise training and inference workloads at scale could progressively displace GPU purchases over a multi-year infrastructure buildout cycle. While hyperscaler in-house chips historically complement rather than fully replace merchant silicon, Alibaba's 20 GW capacity ambition combined with its explicit build-out of domestic and proprietary accelerators adds incremental pressure to Nvidia's enterprise revenue pipeline, particularly in China where export restrictions have already constrained sales.

Jensen Huang has emerged as President Trump's closest adviser on artificial intelligence policy, outflanking tech peers including Mark Zuckerberg and Jeff Bezos for direct influence in Washington. The alignment was on display at the All-In Summit in Los Angeles, where Trump called Huang on speakerphone before the audience — telling him "the robots are not going to be taking over the world" — and Treasury Secretary Scott Bessent later confirmed in a House hearing that "the president is completely aligned with Jensen Huang, the CEO of Nvidia."

Both Trump and Huang share the view that AI safety concerns are overstated and can be addressed through rigorous engineering and pre-release testing rather than new legislation; Trump has called mounting safety warnings a hoax, while Huang has publicly opposed industry-wide R&D slowdowns. Because Nvidia dominates the hardware supply chain behind virtually every major AI model, its CEO's ability to help shape regulatory standards gives the company a structural advantage: rules crafted around Nvidia's testing framework would naturally favor the incumbent chip leader over rivals such as Intel and AMD still struggling to close the gap.