DeepSeek's Disruptive AI Model Sparks Market Volatility and Fuels AI Infrastructure Spending Discussions
On Monday, DeepSeek ignited market turbulence, sending Nvidia (NVDA) shares plummeting over 17%. This development raised concerns about the potential overspending of tech giants like Microsoft (MSFT) and Meta (META) on AI infrastructure.
However, during their respective earnings calls on Wednesday, both Microsoft and Meta executives maintained a positive outlook. They acknowledged the efficiency enhancements introduced by DeepSeek's AI models, but emphasized the ultimate benefits it could bring.
Microsoft CEO Satya Nadella stated that such optimizations will lead to greater AI ubiquity, benefiting hyperscalers and PC platform providers like Microsoft. Meta CEO Mark Zuckerberg played down DeepSeek's threat, suggesting that the industry is constantly evolving and that DeepSeek's announcement is just a temporary setback.
DeepSeek's claim of training its open-source DeepSeek-V3 model for only $5 million on less expensive hardware has sparked concerns about US tech companies being outpaced by Chinese rivals. However, analysts have questioned the accuracy of these claims, highlighting the omission of research and experiment costs.
Zuckerberg acknowledged the uncertainty surrounding DeepSeek's data, but emphasized the importance of continued investment in AI infrastructure. Meta plans to allocate $65 billion towards this initiative in 2023.
Nadella emphasized Microsoft's "fungible" data center strategy, which supports long-term scalability and customer benefits. The company intends to invest $80 billion in AI by fiscal 2025.
Other major tech companies are set to report earnings in the coming days, including Apple (AAPL), Google (GOOG, GOOGL), Amazon (AMZN), and Intel (INTC). Each of these companies is expected to address the current AI market disruption.