Jim Cramer mentioned in the CNBC program last month how and why investors were advised to prune their Nvidia shares. Cramer has long believed in sticking with Nvidia, but has recently changed his point of view, which is a shock for many.

"I wrote an article yesterday and it was very painful for me to write to club members. I made a big idea, it was about the government that you no longer can trust NVIDIA Corp (Nasdaq:NVDA). You can no longer do it. One of the reasons I did this - it turns out that even if I talk about it at the top of the show, the government puts new restrictions on H20, which is the latest NVIDIA CORP (NASDAQ:NVDA), which is my odd number. It happened, it's a different world.

Most of Cramer's focus is related to NVIDIA's Chinese contacts. Now that the United States has reached a 90-day agreement on tariffs with China, it will be interesting to see Kramer's response. Despite this, the core threat to NVIDIA remains.

Nvidia faces challenges on multiple levels. Competition is one of them. Major competitors such as Apple, Qualcomm and AMD are competing for TSMC's 3NM capacity, which could limit Nvidia's access to these chips. Why? Because NVIDIA also uses TSMC's 3NM process node. Nvidia also faces direct competition from other giants who decide to make their own chips. Amazon and its Trainium2 AI chips offer alternatives. The Trainium2 chip provides cost savings and excellent computing power, which could shift AI’s workloads from NVIDIA’s products. Apple is reportedly working with Broadcom to develop AI server processors. Intel is also working to re-enter the game using the Jaguar Shores GPU, which will be produced on its 18A or 14A nodes.

Idaka U.S. growth strategy pointed out the following information regarding NVIDIA Corporation (NASDAQ:NVDA) in its 2025 investor letter in its Q1 Investor Letter:

“Nvidia Corporation (NASDAQ:NVDA) is the undisputed leader in accelerated computing, dominating the graphics processing unit (GPU), delivering AI workloads across data centers, edge devices and emerging platforms. Its end-to-end ecosystem (from silicon to software (CUDA, networking and AI frameworks)) creates higher switching costs and an expanded competitive moat. As secular demand for AI infrastructure remains in its early stages, NVIDIA will benefit from continued Topline growth and strong operational leverage. In early January, DeepSeek, a little-known Chinese artificial intelligence company, released its Big Language Model (LLM) (DeepSeek-R1) to an undesirable world. The model is said to have trained on few high-end NVIDIA chips, which is highly efficient compared to other leading models. The release triggered a ripple effect that investors had to struggle with the idea that the world might not need as many GPUs as previously thought, which hindered NVIDIA purchase cases and reduced P/E multiple times to the cheapest levels in the past 5 years. ”