Taking the world by surprise in 2025, the DeepSeek platform emerged onto the AI scene and caused a stir across global markets. Now, the AI landscape looks vastly different.

What made the Chinese AI app stand out was that its R1 model sought to undercut OpenAI’s ChatGPT – operating powerfully, but at a fraction of the cost.

One year later and DeepSeek services have been heavily adopted across global markets, particularly in Southeast Asia and Africa. Undoubtedly having shaped AI development and interest in 2025, its AI models continue to raise questions about the future of AI development. Companies continue to rethink how much they rely on expensive semiconductors and how high-performance computing (HPC) could evolve moving forward.

How DeepSeek thwarted global technology supremacy

Consequences of the R1 model’s arrival were monumental. Nvidia, whose chips continue to power the most advanced AI systems, had US$600 billion reportedly erased from its value. It was a blow to the US at the time, which had been seeking to slow China’s global AI development with chip export restrictions.

DeepSeek’s progress caught the US technology sector off-guard because it appeared that China could be just as powerful on its own, raising questions over the true impact of technology restrictions. Even US President Donald Trump remarked on the DeepSeek development as a “wake-up call” for US companies.

It also led to the race for AI dominance between China and the US intensifying in 2025, as AI leadership in the West felt threatened.

DeepSeek’s R1 model very quickly became a leading AI model, sparking a change that the most popular models were no longer majority developed in the US. According to Hugging Face’s blog One Year Since the “DeepSeek Moment,” much of 2025’s progress in the open-source community can be traced back to the release of R1.

“DeepSeek’s R1 model lowered the barrier to advanced AI capabilities and offered a clear pattern to follow, unlocking a second layer. Moreover, the release gave Chinese AI development something extremely valuable: time. It showed that even with limited resources, rapid progress was still possible through open source and fast iteration,” Hugging Face’s Adina Yakefu and Irene Solaiman wrote.

“This approach aligned naturally with the goals set out in China’s 2017 “AI+” strategy: combining AI with industry as early as possible, while continuing to build up compute capacity over the long term.”

Can the US stay ahead? Charting global interest in DeepSeek models

US technology companies will not want to be surprised by another ‘DeepSeek moment’. As AI interest and investment continue to surge, 2025 saw significant spending on infrastructure and large private funding for startup companies. In fact, Goldman Sachs predicts that AI companies may invest more than $500 billion in 2026, with major leaders like Microsoft and Nvidia leading capital expenditure (capex).

Amongst this backdrop, the DeepSeek platform keeps growing, with its surge becoming especially attractive in developing regions. This includes a two-to-four-times higher interest in Africa than in other parts of the world, as analysed by Microsoft in its Global AI Adoption in 2025—A Widening Digital Divide report.

In China, the platform now reportedly boasts a market share of nearly 89% among AI users.

2026: AI remains in the spotlight

DeepSeek models continue to disrupt the AI landscape, with The Legal Wire recently reporting that the Chinese government has granted conditional approval for DeepSeek to purchase Nvidia H200 AI chips.

Nvidia CEO Jensen Huang said the company had not yet received official confirmation of these orders, but such a move would follow similar authorisations by ByteDance, Alibaba and Tencent.

The US cleared H200 exports to China in January, but this DeepSeek transaction could spark further geopolitical tensions between the two nations, particularly after Nvidia came under political pressure in Washington after a senior lawmaker said the firm helped DeepSeek optimise its AI systems.

According to a letter, the work focused on squeezing efficiency out of DeepSeek’s training processes, using fewer GPU hours, or computing cycles on specialised chips, to achieve performance rivalrous with leading US models from firms such as OpenAI or Google.

US officials have also previously said DeepSeek’s technology was aiding China’s military, though the company has not publicly confirmed this.

This keeps DeepSeek firmly in the spotlight as the platform prepares to launch its next-generation V4 model this month (February 2026). Although information about V4 is limited, the model is expected to be a coding-optimised model that could push the boundaries of what AI can do for software development.

Reported model testing shows V4 could outperform Anthropic’s Claude 3.5 Sonnet and OpenAI’s GPT-4o on coding benchmarks. However, at the time of writing, this remains unverified.

As the AI race keeps hotting up, what’s clear is that China’s input shouldn’t be underestimated. With infrastructure spending and data centres booming worldwide, China remains active in AI and still operates at the forefront of its development.

Related stories

Cisco CEO: AI is ‘bigger than the internet’, but warns of ‘carnage’ ahead

Marc Ganzi: Why Metro Connect USA matters more than ever in the AI era

ITW Asia: Subsea supply crunch, technology shifts, and the ‘AI question’