<div id="readability-page-1">Nvidia (NVDA 1.74%) stock has tumbled 23% from the record high it reached in early January. The decline began when large language models developed by DeepSeek raised questions about whether U.S. companies have been overinvesting in artificial intelligence infrastructure, especially Nvidia graphics processing units (GPUs). Of course, Nvidia GPUs are the industry standard in accelerating AI applications in the data center. So, any decrease in AI infrastructure spending would be bad news for the company. CEO Jensen Huang addressed those concerns this week during his keynote speech at GTC, an annual developer conference focused on AI. Here's the good news for Nvidia investors. <h2>DeepSeek is not a problem for Nvidia</h2> Chinese start-up DeepSeek shook Wall Street earlier this year when it published technical papers detailing the development of artificial intelligence (AI) models that performed on par with models from U.S. companies like OpenAI and Anthropic. Yet DeepSeek reportedly trained its models with fewer and less powerful Nvidia GPUs, meaning it ultimately spent much less money. CEO Jensen Huang did not mention DeepSeek by name at GTC 2025, but he did answer the question raised by its recent breakthroughs: As more efficient model training techniques bring costs down, will demand for AI accelerators (like Nvidia GPUs) decline? Huang dismissed the idea and said the opposite is true. &#34;This last year, this is where almost the entire world got it wrong,&#34; Huang said. &#34;The amount of computation we need at this point -- as a result of agentic AI, as a result of reasoning -- is easily 100 times more than we thought we needed at this time last year.&#34; Huang says model training efficiencies will actually increase demand for compute capacity by supporting the diffusion of AI into more industries. In other words, more companies can adopt AI as model training costs decline, and the cumulative demand for compute will be at least 100 times more than anticipated before DeepSeek rattled confidence in the chipmaker. Huang impressed his point by comparing GPU shipments among the four largest cloud services providers. Specifically, 3.6 million Blackwell GPUs have been shipped to those companies in 2025 versus 1.3 million Hopper GPUs in 2024. That means collective GPU shipments to the four largest cloud providers -- Amazon, Alphabet, Microsoft, and Oracle -- have increased over 170% in a year despite Blackwell chips being more powerful. Importantly, Huang at GTC also announced the Rubin GPU, the successor to the Blackwell GPU. The Rubin architecture will launch in the second half of 2027, and it will feature 3.3 times more computing power than Blackwell. <div><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MTE3MTAvbnZpZGlhLTEucG5nJnc9NzAw?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" width="700" height="420" original-src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MTE3MTAvbnZpZGlhLTEucG5nJnc9NzAw"/>Image source: Nvidia. </div> <h2>Nvidia is an early leader in the next wave of the AI revolution</h2> Generative AI has been a popular investment theme since ChatGPT launched in late 2022. But the next wave of the artificial intelligence revolution is physical AI, the technology that will enable autonomous cars and robots to understand, navigate, and interact with the real world. Nvidia is well position to be a leader as those nascent markets take shape. &#34;We've been working on self-driving cars now for over a decade,&#34; Huang said at GTC. &#34;We build technology that almost every self-driving car company uses.&#34; Indeed, Tesla leans on Nvidia GPUs in its data centers to train the neural networks that power its full self-driving software. And Alphabet's Waymo not only uses Nvidia GPUs for training in data centers, but also for inference in the robotaxis themselves. Nvidia is the compute platform of choice in autonomous driving systems for two reasons. First, its GPUs are the best data center accelerators on the market. Second, it provides a full-stack solution that brings together hardware and software. Huang explained that at GTC, &#34;We build all three computers: the training computer, the simulation computer, and the robotics computer or self-driving car computer. And the software stack that sits on top of them.&#34; <h2>Nvidia stock looks downright cheap at its current price</h2> Most Wall Street analysts think Nvidia stock is a bargain right now. The median target price is $175 per share. That implies 52% upside from its current share price of $115. To be clear, median refers to the middle value, which means half of the 67 analysts that follow Nvidia expect more than 52% upside in the next 12 months. Building on that, Wall Street estimates the company's earnings will increase 51% in fiscal 2026, which ends in January 2026. That makes the current valuation of 39 times earnings look downright cheap. Those figures give a price-to-earnings-to-growth (PEG) ratio of 0.76. Traditionally, stocks with PEG multiples below 1 are considered undervalued. </div>

NVDA

Nvidia's CEO Jensen Huang reassured investors at GTC 2025, stating that demand for Nvidia GPUs will increase despite concerns raised by the Chinese start-up DeepSeek. Huang emphasized that the need for computation is significantly higher than previously anticipated, driven by advancements in AI. He highlighted a 170% increase in GPU shipments to major cloud providers and announced the upcoming Rubin GPU, which will offer 3.3 times more power than its predecessor. Analysts view Nvidia stock as undervalued, with a median target price of $175, suggesting a 52% upside from its current price of $115.

- Nvidia's stock has dropped 23% since January due to concerns over potential overinvestment in AI infrastructure, particularly in GPUs.  
- CEO Jensen Huang addressed these concerns, asserting that demand for Nvidia GPUs will actually increase as model training efficiencies lower costs, leading to broader AI adoption across industries.  
- Analysts view Nvidia stock as undervalued, with a median target price of $175, suggesting a 52% upside, and projected earnings growth of 51% for fiscal 2026 supports this assessment.

Nvidia Stock Investors Just Got Great News From CEO Jensen Huang