🔥 Nvidia is heading to the next level: 100x more power will be needed for AI!
Nvidia $NVDA continues to crush the competition and its CEO Jensen Huang said at GTC 2025 that the world has underestimated the computing power requirements for AI.
Agent AI = AI that acts autonomously based on user instructions.
Reasoning AI = AI that can logically decompose problems, analyze context and find optimal solutions.
AI needs a huge amount of computation, which means more demand for chips.
"Last year we were all wrong. The computational demands of AI are much higher than we anticipated. Agent and reasoning AI need enormous power," Huang said.
💥 Nvidia vs. DeepSeek
In January, Chinese company DeepSeek shocked the markets when it introduced the R1, which achieves performance comparable to OpenAI and was trained for just $5 million. That's a fraction of the cost compared to the tens of millions of dollars invested in development by US giants.
🚀 DeepSeek V3 - The next generation model that is still cheaper than the US alternatives.
📉 There was panic in the markets for a while because the cheaper AI could reduce demand for extremely powerful chips Nvidia.
But 💪 Huang's response is clear : "The better the AI, the more computing power we need. Competing models are not a problem, on the contrary, they mean even more demand for our chips."
💰 Nvidia plans to invest up to half a trillion dollars in chip production in the US! 🏭
According to Financial Times plans Nvidia a huge investment in chip and electronics manufacturing in the US.
The company plans to invest hundreds of billions of dollars over the next four years. Nvidia does not yet have its own production and is dependent on suppliers (TSMC, Foxconn)but plans to partner with US factories. Competition from China is also forcing Nvidia to strengthen its supply chain in the US.
The Trump administration could accelerate this plan.
🔥 Nvidia faces huge demand for Blackwell AI chips
Huang also confirmed the huge interest in the new Blackwell chips , which will be the basis for the most powerful AI servers.
Major cloud companies have already ordered 3.6 million Blackwell chips, but that's not all, and more companies (Meta, startups, AI companies) are joining in.
Nvidia still dominates the AI market and its chips are best in class.
📉 Potential risks:
If DeepSeek's claim that AI can be trained more cheaply proves true , demand for Nvidia's extremely expensive chips may drop.
However, new regulations and geopolitical risk loom, as governments may begin to restrict the supply of chips to Chinawhich could reduce sales.
Is Nvidia still the best bet for the AI boom, or are more interesting alternatives emerging? What do you think of Nvidia's current pricing?
Nvidia is one of the best companies and with the low prices I unfortunately didn't buy. I still don't find it too cheap so I bought $AMD.
I have so far only bought $AVGO and $ASML from this sector.
Nice overview of new information and news. Nvidia is cheap again and I think the stock is undervalued right now, so I'm buying again after the recent $NVDA sell-off.
The company is still doing well, but the current pricing is quite interesting. I've only been buying $AMD so far, but I may establish a position in $NVDA as well.