DeepSeek versus OpenAI: How the real Artificial Intelligence fight is heating up

“It’s totally hopeless to compete with us on training foundation models,” said Sam Altman shrugging his shoulders dismissively in response to a question on whether a small, smart team with a budget of $10 million could create something substantial in the Artificial Intelligence (AI) space. The question was put to Altman in 2023 during his India visit.

A sobered Altman, on January 28, wrote on X, “it's legit invigorating to have a new competitor!” This was after Chinese DeepSeek triggered a dramatic rethink on artificial intelligence spending around the world, claiming it took just two months and cost under $6 million to build an AI model using Nvidia's less-advanced H800 chips.

Global corporate rivalry is nothing new to the world of business. Nor is the phenomenon of growing US-China tech race for upmanship. Convulsion in the stock markets too is not new. We have had quite a few Nasdaq crashes in the past. However, when DeepSeek, a hitherto unknown tiny Chinese start-up, sent the stocks of US tech giants reeling on 27 December, it is the David vs Goliath novelty that made it fairytale stuff.

The Monday mayhem in the US markets started when the obscure Chinese start-up launched its AI model in the US markets which was rated superior to the existing US models like ChatGPT owned by OpenAI in some respects but developed at a fraction of the cost compared to the US tech majors.

DeepSeek flaunted its model was developed at an unbelievably low cost of $6 million. Compare this with the OpenAI which raised $6.6 billion in the financial markets hardly a few months back to beef up its AI model and the total valuation of the company stood at $157 billion. More than the inclement US weather, this disruptive DeepSeek news sent shivers down the spines of those who had invested in tech stocks. A massive panic withdrawal began.

All US tech stocks put together lost more than $1 trillion in value on that fateful day when the investors got alarmed at the potential competition threat and subversion by DeepSeek. The tech-dominated NASDAQ fell by 3%. The AI chipmaker giant Nvidia, the spearhead of the US tech boom which ranks first among the US tech giants even above Apple, Microsoft and Amazon lost $593 billion, the biggest single-session loss for any company in recent history. Morgan Stanley took the lead among the US rating agencies which rushed to downgrade the target price for many US tech stocks. The stock sell-off was not limited to US alone. Next day, it spread to Europe and Japan as well to turn into a global sell-off.

While tech capitalism has witnessed far greater stock market catastrophes, the unassuming DeepSeek has managed to prick the US tech bubble, giving rise to fears that it might not just be an one-time blip but would set off a series of mini crashes as the US stocks are overvalued indeed. Citi Group analyst Christopher Danely warned that the tech spending by the US companies could slow down unless the tech major showed some comparable return on investment. The Chinese gatecrasher has clearly demonstrated before the world that the apparently mighty US tech giants have feet of clay.

The empire strikes back

The empire did strike back however.

Trump, who had all among championed huge investments in AI by the Silicon Valley majors, was frank enough to admit that DeepSeek-inflicted carnage should come as an eye-opener.

Exactly as they let loose an ultra-nationalist campaign against TikTok, the conservative business lobby alluded to the danger of DeepSeek paving the way for data theft.

DeepSeek reported that it suffered a cyberattack and those who tried to register with it, including from India, were blocked.

The US tech stocks started recovering the very next day. Nvidia again took the lead and rebounded by 2.5. NASDAQ gained 1%, Dow Jones Industrial Average by 0.4% and S&P 500 was up by 0.5%. But the apprehensions lingered. The recovering stocks again witnessed minor fluctuations on Wednesday. The market correction of the US tech stocks might not be over. It is likely to assume the form of a series of mini-fluctuations and corrections.

The euphoria over US tech supremacy

Capitalism works through euphoria. The main underlying reason for the surge of the US tech stocks was the hype that AI is poised to profoundly transform the world. True, AI is the wave of the future and it promises to usher in far-reaching changes in all walks of human life. But the transition would be far more complex and not without corporate institutional hiccups. This is the third blow suffered by the US tech world, especially the pioneer OpenAI.

First, it was the controversy surrounding the firing and the subsequent reinstatement of Altman, the chief architect of AI-driven ChatGPT of OpenAI, as CEO of OpenAI. Altman was also accused of financial irregularities. When the mass of employees of OpenAI, in solidarity with Sam Altman, threatened to resign en masse and he was reinstated. But the episode brought out the lack of transparency and corporate infighting in US tech community.

Second was the controversy surrounding the death of 26-year-old Indian techie Suchir Balajee which was passed off as suicide by the New York Police. But Balajee’s mother alleged it was a murder, citing the ransacking of his room in his San Francisco apartment, unexplained blood stains on his body and the mystery of his missing notes. He had left OpenAI alleging several law violations and other irregularities by the company and he was privy to very sensitive inside information of OpenAI. His mother’s allegation that he was a victim of corporate vested interests evoked broad support among the Indian diaspora tech community.

The DeepSeek assault was the third in a row. But DeepSeek is not the lone disruptive subversive start-up from China. There are also many more waiting to strike in many niche areas like Owen (promoted by Alibaba), MiniMax, Kimi, DuoBao (ByteDance), all from China.

“One important feature the whole DeepSeek episode has categorically demonstrated is that AI has become decentralized. It has spread and proliferated all over the world. The open-source character of its fundamental modules has accelerated this spread. A ‘centralized control’ over its spread by a few corporate biggies is no more possible. Hence, the world must brace itself to embrace the good and evil that unfolds in the future,” professor Murugan, who teaches in Vellore Institute of Technology, The Federal.

The DeelSeek’s foray into the global AI markets has multiple lessons for India too. Contrary to the hype that India has become an AI superpower, India has miles to go before Indian companies could make the kind of wave DeepSeek could whip up in the global AI scenario. True, some Indian IT majors like Tata Elxsi, Wipro, Infosys, Zensar Technologies, and Kellton Tech Solutions and start-ups like Active.ai, Happiest Minds Technologies are already providing AI solutions to data analytics in healthcare and even in manufacture areas like automotive design and tech services. But India is nowhere closer to launching its own version of chatbot embedded in Indian realities, especially regional specificities. A single AI software integrating all the major Indian languages would go a long way in national integration and dampen insular nativism and sectarian irredentism. But it appears to be a long haul. The Indian government has, of course, developed Bhashini, an AI-driven software to real time simultaneous multilingual translation and communication. But a DeepSeek kind of development still remains a far cry.

In the US, Trump has announced a $500 billion federal investment in AI infrastructure and development. But, more importantly, private companies are also chipping in with their contributions. For instance, Meta Platforms are planning to invest $65 billion in 2025. In China, the Bank of China has committed to support AI companies to the tune of $150 billion. There too, Chinese companies like ByteDance is planning to invest $20 billion. In contrast, the size of the Indian AI market is hardly around $8 billion. India has a long way to go.

AI development is inconceivable without supercomputers. Browsing through the list of world’s top 500 supercomputers, we find USA dominates the list with 173 supercomputers, China comes next with 63 supercomputers and India has only 33 supercomputers.

Looking at from another angle, the US has around 60-plus AI labs. Notable among them, besides OpenAI, are Google AI, Microsoft Research AI, Facebook AI Research (FAIR), and AI labs set up in academic centres like MIT-IBM Watson AI Lab, and Stanford AI Lab, Carnegie Mellon University AI Lab etc. In China, there are a whopping 282 AI labs including Alibaba Damo Academy, TenCent AI Lab, Baidu Reaserch and Peking University AI Lab, Tsinghua University AI Lab besides the AI lab set up by the Chinese Academy of Sciences. In contrast, India has only 8 AI labs including those by Google Research India, Tata Consultancy Services, Wipro GE Healthcare and Accenture and the labs at IISc, DRDO, IIIT Bhagalpur and NIEIT. AI research in India needs some major boost both from the government as well as from the corporate titans.

Three reports relating to AI in India

Google in November 2024 and the rating agency Ernst Young (EY) in January 2025 came up with reports laying down the AI route map for India.

The Google report elaborates on how AI in India can improve access to quality healthcare, boost agricultural productivity and sustainability, improve educational and employment opportunities, revolutionize communication amidst India’s linguistic and socio-cultural diversity and improve public service delivery and enhance efficiency in myriad ways in financial sector operations, be it investing in stocks or securities or in lending operations. The Google report also highlights the need for strengthening the digital infrastructure in India that would include cloud computing facilities, improved government databases, and necessary legal frameworks for the IA age. More importantly, the report delves into the most crucial area of preparing an AI-ready domestic workforce through modernizing the labour skilling programs for the AI era and supporting workers in the transition to the new digital AI age. Though the EY report mainly comes up with survey findings on many AI-related parameters in India, it devotes an entire section on what the Indian government needs to do to strengthen AI infrastructure. In other words, it proposes a policy agenda for Indian government. EY report proposes enhancing access to high-quality datasets, developing and enabling access to compute infrastructure, accelerating AI-use case studies/models and R&D, promoting responsible and trusted AI in view of security implications, promoting public-private partnerships, and focusing on sovereign AI capabilities in view of national security considerations.

There is a third related report which is by the International Labour Organisation (ILO) that came out on 28 January 2025. It is not exclusively about India but outlines what governments should do safeguard the interests of labour titled How reskilling for AI could unlock new and better jobs. Strangely, this report doesn’t even pose the question of jobloss compensation by the employers and unemployment allowance by the governments. It focuses only on re-skilling but maintains a total silence on who would bear the cost of re-skilling. So it is left for the labour movement to address the entire gamut of issues thrown up before the labour to cope with the AI transition.