Follow
Follow
Follow
Follow
News
By Rob Enderle for Tech News World
This AI-feeding frenzy reminds me more and more of the dot-com bubble in the late 1990s, but with less focus on overpaying employees. As it was then, the core issue now is companies screaming that their AI is better, but most buyers are still not well enough informed or trained to deploy AI effectively.
One of the big problems DeepSeek seemed to address out of the box is that quality is important in an AI. What makes DeepSeek interesting is that rather than glossing over that problem, it addressed it up front, resulting in what appears to be both a cheaper and better AI alternative.
It is also fascinating that because DeepSeek was blocked from using Nvidia’s current top AI technologies, its developers optimized the model to work better on older Nvidia technology. The result appears to be 94% cost savings in creating the model.
These cost savings have improved performance while reducing expenses. The trade-off is the typical security concern and potential model corruption by the Chinese government, although China may simply be using a filter that blocks unacceptable queries and doesn’t corrupt the core AI.
Let’s discuss AI-wars this week, with a focus on DeepSeek. We’ll close with my Product of the Week from Samsung, which is arguably the most premium non-foldable phone on the market and a showcase of Qualcomm’s latest processor.
Unless you’ve been living under a rock, you’ve heard of DeepSeek. Its entry into the market tanked Nvidia stock last week (even though many argue it shouldn’t have done that) and put OpenAI on notice that there is a new AI sheriff in town from China.
I’ve been arguing for some time that the industry’s huge mistake was focusing more on speed than quality and accuracy because AIs have become increasingly unreliable over time. What DeepSeek did differently was build in an AI quality check. Instead of just focusing on speed, it puts a lot of effort into quality. The result is impressive.
The other advantage to DeepSeek is that while it takes around $100 million to spin up a generative AI model like DeepSeek’s, it claims it did it for only $6 million, resulting in a better, cheaper product. Better and cheaper generally result in a win.
OpenAI is not amused and is suing DeepSeek. However, the lack of cooperation between the U.S. and China makes it unlikely that OpenAI will prevail globally in this effort because China will certainly not support it, which is where DeepSeek is based.
DeepSeek’s origin in China presents a significant drawback because the Chinese government appears to be interfering with the app. Users are blocked from asking questions about anything China finds sensitive or critical of the government’s actions, past or present.
While this looks to be a front-end filter, typically, when you place artificial rules on an AI, it will both seek to break them and foster a knowledge base on how to get around the restrictions, which could eventually result in a degradation of accuracy that defines this product.
Finally, if the U.S. government has issues with TikTok, it will hate DeepSeek because it is more damaging and appears closer to the Chinese government than TikTok. This AI brings up huge domestic privacy and security concerns that easily eclipse those surrounding TikTok, given how an AI is trained and how much more information could be captured from its users.
The competition is just beginning, as we are already up to our necks in new AI companies at 1,532. Most of these companies will be acquired or will fail, which makes this a really interesting and impressively scary market to work or invest in at the moment, much like it was back in those dot-com days when VCs were throwing money at internet companies until it became clear that many of them weren’t going to make it to revenue, let alone profit.
This rapid evolution in AI is only a precursor to what we are likely to see in a few years when the first AGI (artificial general intelligence) products show up. Things will likely get crazy because most of the limitations with generative AI will go away, models should drop rapidly in price, and much like the internet is everywhere now, AI will be on the fast path to being everywhere then.
The amount of money in play is in the trillions of dollars for whoever gets this right first. If Elon Musk is displaced at the top of the rich guy pile, it will likely be by someone as closely connected to AGI as Musk was to electric cars a decade ago. Since AI covers robotics and self-driving vehicles, whoever that is may eclipse Musk quickly once their AGI product sells at scale.
There is already both a fight for top talent and significant efforts to get AI to develop itself. The latter effort is likely to be the most disruptive because once AI can develop itself, the door opens to an even bigger problem with speed and quality unless that quality aspect is addressed as DeepSeek has.
Read the rest of the article to learn Who Will Win the Global AI Race?
About the Author:
Rob Enderle is the president and principal analyst at the Enderle Group, where he provides regional and global companies with guidance on how to create a credible dialogue with the market, target customer needs, create new business opportunities, anticipate technology changes, select vendors and products, and practice zero-dollar marketing. You can reach the author via email.
