@markhmasters You rattled marner when asking him about Berube’s comment about ‘upgrading the play’… what’s up with that Mark? How much you got on the panthers?
#DeepSeek R1 was trained on over 2,000 Nvidia GPUs, which is an impressive technical foundation. However, while the headlines are busy celebrating this so called game changer from the Chinese AI space, it’s worth taking a step back to look at the bigger picture. Artificial intelligence is not just a toy for language models. It’s the driving force behind the next industrial and technological revolution.
Facial recognition, medical diagnostics, autonomous systems, and even platforms like Palantir rely on one essential element, massive computational power. Even if DeepSeek R1 has indeed been trained on cheaper Nvidia H800 chips, that’s merely an optimized interim solution. The future isn’t about training models cheaply. It’s about deploying those models at scale and making them work efficiently in real-time.
Take autonomous driving, for example. It’s not enough to have an AI model that can theoretically analyze traffic. That analysis needs to happen in milliseconds, processing billions of data points simultaneously. The same applies to medical AI applications, where X-rays, MRIs, and patient data must be evaluated instantly. All of this requires GPUs or other specialized hardware.
DeepSeek R1 is being positioned as a cost-effective alternative to models like ChatGPT or GPT-4, but at the end of the day, it’s a response to U.S. export restrictions on China. The current hype surrounding this “breakthrough” should be viewed with some skepticism. This is a strategic reaction to economic and geopolitical constraints, and the underlying data and technology aren’t always transparent enough for an objective assessment.
For investors or tech enthusiasts, it’s crucial to analyze this hype rationally. Even if private users start installing DeepSeek at home, they’ll still need GPUs—and those are made by #Nvidia or #AMD. These companies remain the backbone of AI infrastructure. The idea that a single model like DeepSeek could destabilize established industry leaders is simply unrealistic.
I tested the new large language model myself last night. While it performs well in some areas, it doesn’t match ChatGPT in its latest version, especially in fields like forensic anthropology. The quality of responses isn’t on the same level. This isn’t surprising. China doesn’t always tell the full story, and we shouldn’t take every claim at face value.
For Nvidia shareholders, today is an opportunity to pick up shares at a discount. The focus shouldn’t be on the next six months but on the next ten years. The demand for computational power will only grow as AI evolves. Models like DeepSeek R1 are merely early milestones in a journey that will require exponential growth in hardware capacity. Those who can see the long game will realize that Nvidia and similar companies aren’t under threat. They are essential pillars of a future dominated by artificial intelligence. Be cool, and don’t panic.