DeepSeek’s sudden rise was a shock to Silicon Valley — a cheap-to-build Chinese chatbot rivaling ChatGPT wiped $600bn off Nvidia’s value in one day and rattled US tech dominance.
Its low-cost model challenged the belief that only huge budgets can produce top AI, making it a potential “Sputnik moment” for the AI race between the US and China.
#1Six months on, DeepSeek is no longer dominating headlines, but it’s still quietly shaping the AI landscape.
The Chinese chatbot upended the “bigger is better” mindset in Silicon Valley, proving that smart engineering can rival massive data centers and expensive hardware. Despite security concerns and corporate bans, many US startups still use it to cut costs—some even running it locally to avoid sending data to China. For cash-strapped firms, the savings often fund more critical needs like hiring.
#2DeepSeek’s debut marked a major turning point in the US–China AI rivalry.
Once seen as lagging behind Western AI, China stunned the tech world when DeepSeek claimed it had built a top-tier large language model—outperforming OpenAI’s o1 on multiple benchmarks—using just $5.6 million, compared to OpenAI’s $5 billion spend in 2024. The feat shifted perceptions of China’s AI capabilities and pushed US policymakers and tech giants to frame AI development as a matter of national security.
But DeepSeek’s Chinese origins continue to raise red flags in Washington. US officials allege the firm could aid China’s military and intelligence operations, while its privacy policy confirms all user data is stored on servers inside China. The company has declined to comment on these concerns.
#3Earlier this week
, OpenAI stirred fresh comparisons to DeepSeek by releasing its first free, open-source AI models in five years. Industry voices like d-Matrix’s Sid Sheth see a direct link, crediting DeepSeek for proving smaller, cheaper models can still perform at a high level—spurring a shift toward “right-sized” AI.
Yet, the big-budget arms race is far from over. Days after the open models, OpenAI launched GPT-5, backed by massive new computing capacity. Meta is spending billions and offering $100M pay packages to lure talent, while Nvidia’s stock has rebounded to record highs.
The initial “DeepSeek disruption” has faded. The industry is once again betting on more chips, more data centres, and more power. Meanwhile, DeepSeek itself is struggling to keep momentum—facing stiff competition and delays to its next model, DeepSeek-R2, partly due to a shortage of high-end chips.
Reactions
Already reacted for this post.
