With its low-cost AI model, DeepSeek, a Chinese AI company, has sent shockwaves through the U.S. tech market. The company’s chatbot has quickly become the most downloaded free app on Apple’s U.S. app store, surpassing OpenAI’s ChatGPT. This development has major implications for Silicon Valley.
Expert @_KarenHao on what Deepseek may mean for the future of journalism and Gen AI.
"DeepSeek will be very inspirational for AI researchers who now realise they can develop approaches that require significantly less money."
✍️ A piece by @Marina__Adamihttps://t.co/yoUP9oZiP7
— Reuters Institute (@risj_oxford) February 1, 2025
“While AI technology has provided hugely important tools… the business model depends on hype.”
Excellent piece on the significance of DeepSeek by @kenanmalik https://t.co/2YFyWRBr8E
— John Tasioulas (@JTasioulas) February 2, 2025
DeepSeek’s cost-effective models directly threaten the expensive AI technologies funded by American tech giants like Google, Microsoft, Apple, and Meta. These companies have invested billions into advanced AI, sparking an arms race in the industry. On Monday, the market reacted to the rise of DeepSeek.
“While privacy fears are justified, the main beef Silicon Valley has is that China’s chatbot is democratising the technology“
[DeepSeek has ripped away AI’s veil of mystique. That’s the real reason the tech bros fear it | Kenan Malik | The Guardian] https://t.co/gxZyqf80yC
— Norio Nakatsuji (@norionakatsuji) February 2, 2025
Nvidia, a leading chip company known for its AI technology, saw its stock drop more than 13% by late morning. Other chip companies, including Arm and Broadcom, also experienced significant declines. By late morning, the tech-heavy Nasdaq fell almost 600 points, or nearly 3%.
Investors are concerned about the long-term sustainability of AI investments. DeepSeek claims it can use less advanced chips to train its models at a lower cost. This questions the competitiveness of the hefty investments made by Big Tech.
“The AI startup has upended the industry by developing a model that costs much less to produce – and is available free to a universe of tinkerers“
[DeepSeek: cheap, powerful Chinese AI for all. What could possibly go wrong? | John Naughton | The Guardian] https://t.co/HcRyHfyVyb
— Norio Nakatsuji (@norionakatsuji) February 1, 2025
In a critical week for the tech sector, four major companies—Meta, Apple, Microsoft, and Tesla—are set to report their quarterly results. Investors are eager for insights on how these companies plan to respond to DeepSeek’s rise and adjust their AI strategies.
DeepSeek challenges U.S. tech giants
Nvidia quickly responded to the shock caused by DeepSeek’s announcement. The company incorporated the DeepSeek R1 model into its Nvidia NIM microservice, which will soon be fully integrated into the Nvidia Enterprise software platform. This move allows developers to access AI models and build applications, potentially driving up the demand for Nvidia’s latest GPUs.
Nvidia also highlighted the importance of high-powered tools for AI inferencing. The company stated, “Inference requires significant numbers of Nvidia GPUs and high-performance networking.” This indicates that despite the performance of lower-powered chips, Nvidia’s premium GPUs remain essential for top-tier results. DeepSeek-R1 is an open model with state-of-the-art reasoning capabilities.
It performs multiple inference passes over a query, using chain-of-thought, consensus, and search methods to generate the best answer. This process, known as test-time scaling, requires significant computing power for real-time inference and higher-quality outputs. The DeepSeek-R1 microservice can deliver up to 3,872 tokens per second on a single NVIDIA HGX H200 system.
Developers can test and experiment with the API, which will soon be available. The microservice simplifies deployments, supporting industry-standard APIs and allowing enterprises to maximize security and data privacy by running it on their preferred accelerated computing infrastructure. Nvidia’s latest architecture will significantly boost test-time scaling on reasoning models like DeepSeek-R1.
The fifth-generation Tensor Cores can deliver up to 20 petaflops of peak FP4 compute performance and a 72-GPU NVLink domain optimized explicitly for inference. Developers can now experience the power of DeepSeek-R1 through the microservice. With these tools, enterprises can deploy the model effortlessly, ensuring they achieve the high efficiency needed for agentic AI systems.