Chinese AI DeepSeek R1 disrupts tech industry with low-cost efficiency

(CNN) — A new player in the artificial intelligence landscape is causing quite a stir. The Chinese-developed AI model, Deepseek R1, has made waves in the tech industry and is even drawing attention on Wall Street. More about its effects on the market can be read here.

DeepSeek, a startup that emerged only a year ago, developed this innovative AI model. Remarkably, R1 is proving to be nearly as capable as well-known AI models such as OpenAI's GPT-4, Meta's Llama, and Google's Gemini, but at a fraction of their cost. Marc Andreessen, a renowned tech investor, described this development as “AI’s Sputnik moment.”

With an investment of only $5.6 million in training the R1 model, DeepSeek has managed a significant feat, especially considering American firms spend hundreds of millions, if not billions, on similar AI technologies. This is particularly noteworthy given that the United States has attempted to limit China's access to high-performance AI chips due to security concerns.

DeepSeek’s Origins

Founded in late 2023 by Chinese hedge fund manager Liang Wenfeng, DeepSeek is among numerous startups aiming to capitalize on the AI boom. Liang has been likened to Sam Altman of the U.S., advocating for AI innovation and funding through his hedge fund, High-Flyer.

Following in the footsteps of other AI startups like Anthropic and Perplexity, DeepSeek has released multiple AI models over the past year. Their V3 model garnered some interest despite its limitations regarding sensitive topics pertaining to the Chinese government, as reported by the Wall Street Journal.

The unveiling of R1, which happened last year, caught everyone by surprise, especially with its low operational costs. The model is open-source, allowing other companies to refine and build upon it. The DeepSeek app has quickly climbed the app store rankings, surpassing ChatGPT, with close to 2 million downloads.

Implications of DeepSeek’s Innovation

Artificial intelligence is known for being resource-intensive, both in terms of power and cost. For instance, some leading tech firms in the U.S. are investing in nuclear power to meet their AI models' energy requirements, detailed further here.

Meta recently announced plans to invest over $65 billion in AI this year. OpenAI CEO Sam Altman previously indicated the necessity of trillions of dollars for AI chip advancements to support data centers powering such technologies.

The fact that DeepSeek can achieve comparable AI capabilities at a fraction of the cost challenges the industry's current financial assumptions. While there are critics of AI, its proponents argue for a transformative economic future driven by AI, offering new efficiencies and opportunities.

Andreessen, who co-founded the venture capital firm Andreessen Horowitz, praised DeepSeek as “one of the most amazing and impressive breakthroughs I’ve ever seen,” in a post on X.

Impact on the U.S. Tech Industry

The United States has attempted to maintain technological supremacy by restricting exports of crucial AI components. Just before leaving office, former President Joe Biden reinforced these measures, aiming to prevent countries like China from obtaining advanced AI technology.

DeepSeek’s achievement challenges this strategy and raises questions about the United States’ dominance in the tech industry. Despite trade restrictions, America's edge in AI has diminished.

This development has sparked concern on Wall Street. U.S. stocks faced a decline, with Nvidia (NVDA), a major supplier of AI chips, seeing a 12% drop in premarket trading. Other major tech stocks like Meta (META) and Alphabet (GOOGL) were also affected.

Evaluating the Long-term Significance

The industry largely believes in the authenticity of DeepSeek's cost claims, although skepticism remains due to the company's relative obscurity. Despite the dramatic market reactions, it's too soon to dismiss U.S. tech leadership entirely.

Keith Lerner, an analyst at Truist, remarked, “The DeepSeek model rollout is leading investors to question the lead that U.S. companies have and how much is being spent and whether that spending will lead to profits (or overspending). Ultimately, our view, is the required spend for data and such in AI will be significant, and U.S. companies remain leaders.”

While DeepSeek’s cost efficiency is notable, its R1 model primarily targets consumer applications like ChatGPT. Its capability for more significant AI functions requiring large-scale infrastructure investments remains to be proven.

Giuseppe Sette, President of AI market research firm Reflexivity, expressed, “Thanks to its rich talent and capital base, the US remains the most promising ‘home turf’ from which we expect to see the emergence of the first self-improving AI.”