Wall Street’s Unwavering Faith: Analysts Maintain Buy Ratings on Nvidia Despite Valuation Fears and Rising Competition
Wall Street isn't blinking. Across the board, analysts are holding their 'Buy' ratings on Nvidia, even as the specter of sky-high valuations and a growing pack of rivals looms large.
The Bull Case: Built to Last
Forget the noise about competition heating up. The consensus view from the Street suggests Nvidia's moat is deeper than the doubters think. Its architecture isn't just hardware—it's the de facto operating system for the AI revolution, creating a sticky ecosystem that's brutally hard to bypass.
Valuation? What Valuation?
Sure, the stock trades at a premium that would make a traditional value investor's head spin. But in the land of AI, old metrics get tossed out the window. The bet here is on sheer dominance in a market that's still being defined—a classic 'winner-takes-most' scenario playing out in real-time. It's the kind of math that only works until it doesn't, but for now, the Street is all in.
The Hedge Fund Whisper
Let's be real—keeping a Buy rating is the easy part. It's the price target whispers behind closed doors that tell the real story. No one gets fired for owning the leader, even if they secretly think the easy money has been made. It's the ultimate 'cover your assets' move in a game of momentum and narrative.
The bottom line? Wall Street is placing a massive, unified bet that Nvidia's engine has enough fuel to power through the turbulence ahead. They're betting the house that in the race for AI supremacy, there's still only one name that truly matters.
Bank of America pushes its case after new meeting details
Bank of America’s Vivek Arya met with Toshiya Hari, who handles investor relations at the company, and doubled down with a $275 target. He kept the buy rating and said the stock remains a top pick. He laid out several points from the meeting.
One of them was that the company agrees that Google’s Gemini 3 is a strong LLM trained on TPUs, but it said it is too early to call a winner. Management reminded him that all existing GPU models on the market today rely on the older Hopper design from 2022.
Those models cannot be compared with what is coming next, because the next wave uses the 2024 Blackwell chips. That difference matters for training and inference.
Management told Arya they expect the first Blackwell-backed LLMs to arrive in early 2026 and said the new chips should put the company “at least a full generation ahead of competition.” External benchmarks like MLPerf and InferenceMAX also list Blackwell at the top for tokens per watt and revenue per token.
Arya added that the chip Maker has visibility into at least $500 billion of combined demand and supply across Blackwell, Rubin, and networking for 2025 and 2026. He said the recent letters of intent with OpenAI and Anthropic/Microsoft are not even included in that $500 billion number and could add more.
Arya pointed to valuation data, too. He said the company trades at 25x and 19x its 2026 and 2027 earnings, which gives a PEG ratio of 0.5x. He compared that with the 2x average for the Magnificent Seven and other growth names. He ranks No. 270 on TipRanks, with profitable calls 58% of the time and an average return of 17.7%.
Bernstein and Jefferies build on upside expectations
Bernstein’s Stacy Rasgon also kept a buy call and a $275 target after his VIRTUAL meeting with Stewart Stecker from the investor relations team.
Rasgon said the $500 billion outlook for Blackwell, Rubin, and networking revenue for 2025 and 2026 could see upside because it did not include deals with Anthropic, the OpenAI 10-gigawatt collaboration, or work underway in the Middle East.
On competition from Google, he said the company believes it sits about two years ahead of Google’s TPU program. He added that the team thinks cloud providers will be slow to adopt TPUs because they work best with certain model structures.
He included a comment from the meeting: “But they believe NVIDIA’s programmable platform solutions remain the best hardware for cloud AI infrastructure.”
Rasgon also addressed President Donald Trump’s post about allowing shipments of H200 chips to China with a 25% cut to the U.S. The company is still waiting on licenses before manufacturing and has not received details on how that revenue share WOULD be handled. Rasgon ranks No. 144 on TipRanks, with a 67% success rate and an average return of 27.3%.
Jefferies analyst Blayne Curtis stayed bullish with a $250 target while calling Broadcom his top pick because of expected ASIC momentum.
Still, he said he is not stepping away from Nvidia. His note said, “We haven’t given up on NVDA given the technology moat and valuation at 18x the $10 EPS bogey.” Curtis said ASIC adoption is early and leaves room for growth as spending rises.
He said worries are overstated because Blackwell Ultra is on track and Rubin should ramp in late 2026. He also expects Vera-Rubin and NVLink 6 launches in the second half of 2026. He thinks Blackwell-based LLMs coming in the first half of 2026 could help the stock.
Curtis added that the CPX chip launching in late 2026 could benefit from higher hyperscaler spending and more demand for inference.
Curtis expects CPX revenue of $13 billion in 2027 and raised his EPS forecasts for 2026 and 2027 to $7.82 and $9.50. He ranks No. 58 on TipRanks, with calls that have been profitable 64% of the time and an average return of 27.8%.
Get $50 free to trade crypto when you sign up to Bybit now
