What is DeepSeek, the Chinese AI startup shaking up the tech world?

An unexpectedly potent and effective Chinese AI model has swept the tech sector. It is causing anxiety on Wall Street and is known as DeepSeek R1.

Deepseek

The new AI model was created by DeepSeek, a startup that was only founded a year ago. Somehow, the company has achieved what renowned tech investor Marc Andreessen has dubbed "AI's Sputnik moment": R1 can almost match the capabilities of its much more well-known competitors, such as Google's Gemini, OpenAI's GPT-4, and Meta's Llama, but at a much lower cost.

In contrast to the hundreds of millions, if not billions, of dollars that US firms spend on their AI technologies, the company claimed to have only spent $5.6 million to fuel their main AI model. Given that the US has been trying for years to limit China's access to high-power AI chips due to national security concerns, it is even more startling. This indicates that DeepSeek allegedly succeeded in achieving its low-cost model on AI processors with comparatively little processing power.

What is DeepSeek?

The business was established in late 2023 by Liang Wenfeng, a Chinese hedge fund manager. It is one of several businesses that have emerged in recent years looking for significant funding to capitalize on the enormous AI wave that has propelled the tech sector to unprecedented heights.

Liang has turned into China's equivalent of Sam Altman, a proponent of AI technology and funding for fresh studies. AI development is the main focus of his hedge fund, High-Flyer.

Similar to other AI startups such as Anthropic and Perplexity, DeepSeek has introduced several competitive AI models in the past year, garnering attention within the industry. The release of its V3 model brought some recognition to the company; however, the implementation of content restrictions concerning sensitive subjects related to the Chinese government and its leadership has raised questions regarding its potential as a viable competitor in the market, as reported by the Wall Street Journal.

R1, which emerged unexpectedly when it was announced late last year, was launched last week and garnered considerable attention this week after the company disclosed its remarkably low operational costs to the Journal. Additionally, it is open-source, allowing other companies to experiment with and enhance the model.

The DeepSeek application has rapidly ascended the app store rankings, overtaking ChatGPT on Monday, and has achieved nearly 2 million downloads.

What makes DeepSeek so significant?

AI represents a technology that demands significant energy and financial resources. This has led some of the most influential technology executives in the United States to acquire nuclear power companies to ensure a stable electricity supply for their AI systems.

Recently, Meta announced plans to invest over $65 billion in AI development this year. Additionally, Sam Altman, the CEO of OpenAI, indicated last year that the AI sector would require trillions of dollars in investments to facilitate the creation of the highly sought-after chips essential for powering the energy-intensive data centers that support the intricate models used in the industry.

The idea that comparable functionalities to those of America's leading AI models can be realized at a significantly lower cost and on less advanced hardware signifies a fundamental shift in the industry's perception of the necessary investment in artificial intelligence. While the technology faces considerable skepticism and opposition, its proponents assert a promising outlook: they contend that AI will propel the global economy into a new phase, enhancing efficiency in the workforce and unlocking new potentials across various sectors, thereby facilitating innovative research and development.

What implications does this hold for the United States?

The United States believed that imposing sanctions would secure its supremacy in a crucial technology deemed essential for enhancing national security. Merely a week prior to his departure from office, former President Joe Biden reinforced export limitations on AI computer chips to hinder competitors such as China from obtaining this advanced technology.

However, DeepSeek has challenged this perspective, undermining the perception of invulnerability associated with America’s technology sector. While the U.S. may have temporarily delayed its rivals through these export restrictions, its advantage in AI has significantly diminished in light of these measures.

DeepSeek may indicate that disabling access to a crucial technology does not guarantee a victory for the United States. This serves as a significant message for President Donald Trump as he continues to implement his isolationist "America First" agenda.

The development has caused considerable concern on Wall Street. US stocks were poised for a significant decline on Monday morning. Nvidia (NVDA), the foremost provider of AI chips, which has seen its stock price more than double over the past two years, experienced a 12% drop in premarket trading. Similarly, shares of Meta (META) and Alphabet (GOOGL), the parent company of Google, also saw substantial declines, along with other major technology firms such as Marvell, Broadcom, Palantir, and Oracle.

Are we truly certain this is significant?

The industry is believing the company when it claims that the expense was so minimal. No one is truly arguing against it, but the market panic relies on the honesty of one specific and somewhat obscure company. The company notably omitted the expense incurred in training its model, leaving out possibly high research and development expenses. (However, it likely didn’t invest billions of dollars.)

It's still much too early to dismiss American technological innovation and leadership. A single accomplishment, impressive as it may be, might not be enough to offset years off advancements in American AI dominance. A significant customer transition to a Chinese startup is improbable. 

While the cost-saving accomplishment may be considerable, the R1 model serves as a rival to ChatGPT — a large language model aimed at consumers. It has not yet demonstrated the ability to manage certain highly ambitious AI functionalities for sectors that — at this moment — still need significant infrastructure investments.

Disclaimer : Above all information taken from multiple source, newsandblogs cannot be responsible held for any errors or omissions.