DeepSeek: “Let A Hundred Schools of Thought Contend!”

Only a week after American oligarchs flaunted their access to the White House, we could have expected their share prices to be on a high. However, since last week, the message coming from Silicon Valley is far less rosy as Wall Street tech stocks are set to lose over a trillion dollars in value. This crisis of confidence in what was until recently considered an insurmountable monopoly backed by a global superpower originates with a hitherto unknown start-up.

DeepSeek, a Chinese company founded in 2023, has evidently dethroned OpenAI as the world’s premier AI venture. This comes only a few days prior to President Trump coronating OpenAI as the leading developer in his administration’s ambitious AI development programme dubbed Project Stargate. This commitment to investing hundreds of billions into AI infrastructure was at one point taken to herald a new gilded age for Silicon Valley and further consolidate the existing American dominance in this critical emerging sector, well, at least this was believed to be the case until DeepSeek arrived on the scene. In fact, since President Trump’s inauguration, China has already approved projects to build the world’s largest electrical dam, released the world’s fastest train, started testing 6G internet faster than Elon Musk’s infamous Starlink and started constructing a kilometre-wide solar space farm in our atmosphere with an efficacy tenfold higher than terrestrial farms so it is without surprise that DeepSeek is building upon this litany of innovation under President Xi’s dynamic tenure.

DeepSeek continues to be the most sophisticated and threatening rival to OpenAI’s current ChatGPT with DeepSeek’s R1 operating system mercilessly outperforming OpenAI’s O1 system at comparable stages of development. Additionally, DeepSeek continues to build equally impressive multimodal AI models with Janus Pro 1B and 7B able to outperform OpenAI’s DALL·E 3 and Stability AI’s Stable Diffusion. Such feats have translated to the top charts of mobile app stores, unseating OpenAI’s ChatGPT app as DeepSeek continues to conquer an undersaturated market. To summarise, DeepSeek has not only started the AI race, DeepSeek has unabashedly reshaped the entire landscape to the simmering ire of Washington.

Besides, as if to administer salt over the open wound to which OpenAI’s defunct fiefdom has been reduced, DeepSeek was able to achieve the above in a far more efficient and benevolent manner. The dizzying heights to which DeepSeek has soared thus far have been reached on a humble budget of $6,000,000, a modest sum when compared to the multi-billions OpenAI continues to demand from the American people. Moreover, this broader achievement should be re-contextualised in terms of the ongoing export controls unleashed on China which only enabled NVIDIA H800 accelerators to be used by DeepSeek as opposed to the superior H100 accelerators freely accessible to American developers. Nevertheless, the bubble has burst and seems to be taking the entire sector with it. NVIDIA, now a former darling of Wall Street, has already lost over $600,000,000,000 in market cap and fallen near 20% due to the simple reality that DeepSeek was able to outperform OpenAI on a modicum of the budget and a fraction of the semiconductors. Officially, NVIDIA has experienced the biggest market cap loss in 24 hours in the history of the stock market, equally, striking NASDAQ and S&P 500 indexes into capitulation and closure. Such developments will raise important questions for Washington over the efficacy of crude controls towards containing China’s economic miracle.

Finally, the undeniable pièce de résistance of DeepSeek’s resplendent florescence remains the fact that it is open source. OpenAI, contrary to the name, is not and has never been open source, despite this stated goal in its founding mission. Instead, OpenAI’s CEO Sam Altman, similarly to other tech giants, seems intent on a two-pronged approach of centralising AI development while concurrently exercising direct influence over shaping AI policy. OpenAI, founded as a non-profit in 2015, aimed to develop AI that would benefit “all of humanity” but would end up transitioning to a “capped-profit” model in 2019 in order to raise capital from established investors like Microsoft. This move marked a significant departure from earlier commitments to open and accessible technology. For example, while earlier models like GPT-2 were released alongside detailed papers and model weights, subsequent models like GPT-3 and GPT-4 have been accompanied by increasingly limited transparency over architecture, training data and risk. This receding accountability, compounded by Altman’s own efforts to advocate and lobby for regulation that may stifle smaller competitors and open-source initiatives, paints a grim image as OpenAI continues to shape AI policy for regulatory bodies, internationally. This all follows reports from OpenSecrets detailing the exponential expenditure from OpenAI. As expected, this shameless wedding between state and monopoly interests can only be to the disadvantage of the populations captivated by AI technology as attempts to preserve monopoly continue to retard development. Ultimately, while OpenAI maintains an infrastructural advantage and a loyal user base, as things currently stand, the uncontested domination which they have come to cherish over the last few years is undeniably coming to an end as a new epoch for AI unfolds before our own eyes.

As we enter the Year of the Snake, myself among others rejoice that the People’s Republic of China will remain the paramount thorn in the side of our inhuman oligarchy intent on curbing innovation and prosperity for those within the West and beyond, may a hundred flowers bloom and may a hundred schools of thought contend!