Enter the dragon! Will DeepSeek upend the US AI tech hegemony? Part - 1

“OpenAI is not a god; they won’t necessarily always be at the forefront.”

These are the words of Liang Wenfeng, Founder and CEO of DeepSeek, the new kid on the AI block that is shaking up the tech world. Though this Chinese AI start-up has been in the game for a while, it is their latest breakthrough – the R1 model – that is stealing the spotlight. Why? Because the open-source model’s performance rivals that of ChatGPT in many third-party tests, making waves across the industry.

Wall Street went into turmoil, with leading chipmaker Nvidia’s shares plummeting as much as 18%, the biggest market value drop in US stock market history, according to Bloomberg. The shake-up highlights investor anxiety as US tech giants struggle to balance aggressive AI investments with profitability.

The mysterious force from the East

Founded in 2023 and headquartered in Hangzhou, home to tech giants like Alibaba, DeepSeek quickly earned the moniker of Silicon Valley’s “mysterious force from the East”. Unlike most AI firms, it has resisted commercialization, relying solely on internal funding—a rarity in today’s tech race.

“Money isn’t our problem,” Wenfeng states. “US chip bans are.” He dismisses the notion that bigger investments guarantee innovation, arguing, “If that were true, big firms would’ve monopolized all breakthroughs by now.”

DeepSeek’s philosophy is deeply rooted in open-source development, which Wenfeng describes as a “cultural behaviour rather than a commercial strategy.” Unlike competitors that pivot to closed-source models, DeepSeek remains committed to transparency and collaboration.

Liang Wenfeng stresses that the real gap in AI is not just in time, but in innovation culture. While Western tech firms have built robust ecosystems over decades, China lacks a similarly collaborative and knowledge-sharing environment.

A different approach to AI

While the leading AI companies rely on supercomputers with up to 16,000 chips or more to train their chatbots, DeepSeek claims it only required around 2,000 Nvidia chips to develop its AI model.

What truly sets the DeepSeek’s R1 model apart is its ground-breaking architecture, designed to optimize memory usage. This means DeepSeek operates with significantly less computational power compared to resource-intensive market leaders like ChatGPT, redefining efficiency in AI performance.

It’s no small feat, especially considering the US sanctions on advanced chip imports to China. And it is touted as a countermove to the US, particularly in the context of the TikTok ban.

DeepSeek claims it developed the R1 model in just two months for under $6 million – 10 times less than Meta’s latest AI project. If true, this challenges the belief that AI progress depends on massive financial backing. However, skepticism remains. Analysts like Stacy Rasgon of Bernstein and Ralph Schackart of William Blair question whether DeepSeek’s rapid advancements are as “miraculous” as they appear.

The future of AI: A global shift?

It is a fact that this Chinese start-up has signalled a bold challenge to the AI status quo. Venture capitalist Marc Andreessen referred to DeepSeek’s R1 model as “AI’s Sputnik moment”, likening it to the Soviet Union’s 1957 satellite launch that triggered the Cold War space race with the US.

As DeepSeek disrupts the AI hierarchy, key questions emerge: Will China outfox the US in the AI arms race? Or are we on the brink of witnessing the true democratization of AI? In this two-part series Adgully seeks to find answers with insights from tech experts.

Bruce Keith, Co-Founder & CEO of InvestorAi, believes that this development “genuinely democratizes AI,” giving countries without existing infrastructure a significant advantage to experiment and participate at the frontier of innovation.

“DeepSeek R1 has undeniably challenged the dominance of key players in the models and data ecosystem – OpenAI, Google, and Meta will feel it the most. R1 will have a profound impact on the AI landscape, emphasizing the importance of innovation and focusing on applications and data, rather than solely on processing power. The fact that this platform was created with under $6 million in investments has sent shockwaves through the tech world, proving that game-changing innovation doesn’t always require billion-dollar budgets. However, from a supply-and-demand perspective, the GPU market, which Nvidia dominates, is still far from reaching peak demand,” he adds.

Abhinav Jain, Co-Founder and CEO of Almonds AI, is unequivocal in his belief that DeepSeek AI’s success with the R1-Zero model is a game-changer.

According to Jain, seeing a relatively new player like DeepSeek rise to challenge industry giants like OpenAI and Google is a wake-up call for the entire AI sector.

“What’s remarkable is how quickly DeepSeek has made an impact. The R1-Zero model has not only matched but, in some cases, exceeded the performance of OpenAI’s models, all while being significantly more cost-effective. The fact that it is now the most downloaded app in the US speaks volumes about its quality and appeal. The $600 billion dip in Nvidia’s market cap after DeepSeek’s launch is telling. It shows just how disruptive this breakthrough is, especially when you consider that DeepSeek achieved this in the face of US sanctions on advanced chip imports. How did they do it? Well, first, they embraced open-source AI models and found innovative, low-cost ways to run their computations. This wasn’t just a clever workaround; it was a bold statement that you don’t need sky-high budgets to create world-class AI,” adds Jain.

Jain applauds the founder Liang Wenfeng, whose vision and experience are key drivers here. He observes that Wenfeng is committed to making AI more accessible and affordable, and that mission resonates in every aspect of DeepSeek’s work.

According to him, their use of distillation techniques is brilliant – essentially packing the knowledge and power of massive models into smaller, more efficient ones. It’s not just about competing with the giants; it is about democratizing AI for a much broader audience.

The launch of DeepSeek AI has drawn significant attention in the US tech landscape, challenging the need for multi-billion-dollar investments in AI models when China has demonstrated the ability to develop similar technology at a fraction of the cost, says Alankar Saxena, Co-founder and CTO of Mudrex.

Notably, he points out, the launch comes shortly after President Trump announced a $500 billion AI infrastructure investment, signalling China’s intent to compete directly with the US.

“This launch sets a new standard for the tech giants, prompting them to rethink the high spending on building AI models. While both nations still have substantial progress to make, this rivalry could ultimately benefit consumers by driving innovation, improving services, and creating more competitive pricing in the AI market,” Saxena says.

The global AI landscape is shifting with the emergence of DeepSeek AI, whose R1 model was developed for under $6 million – far less than the hundreds of millions typically spent by competitors, points out Giridhar LV, Founder, Nuvepro Technologies. Its efficiency, achieved through innovative techniques like reinforcement learning without traditional fine-tuning, raises questions about the sustainability of massive investments by US tech giants.

He believes that despite US restrictions on chip exports, China has shown resilience, using creativity and collaboration to overcome hardware limitations. “This challenges US dominance in AI and signals a shift in global tech power. DeepSeek’s rise emphasizes the growing importance of efficiency and innovation over sheer financial resources, signalling a more competitive and dynamic global AI market moving forward. DeepSeek’s success has been described by former US President Donald Trump as a “wake-up call” for American tech, highlighting the potential benefits of innovation at lower costs, though it has also raised cybersecurity concerns, especially in countries like Australia,” he adds.

(Tomorrow: With DeepSeek boasting lower training costs than US rivals, could this reshape global AI innovation? As China’s AI sector gains ground, what geopolitical or economic shifts might follow? Part 2 of this report will dwell on these questions and more.)

Also Read: Bruce Lee's - 'Enter The Dragon' on 03rd June at 8 PM