Following Chinese startup DeepSeek’s announcement of an AI model developed at a fraction of typical industry costs, US tech leaders Microsoft and Meta have reiterated their commitment to substantial AI infrastructure investments. DeepSeek’s model, reportedly developed for under $6 million, has prompted discussions about the efficiency of large-scale AI expenditures.
Meta CEO Mark Zuckerberg stressed the importance of significant capital expenditure, stating, “Investing ‘very heavily’ in capital expenditure and infrastructure is going to be a strategic advantage over time.” Meta plans to allocate between $60 billion and $65 billion to AI development this year, including constructing a substantial data centre.
Microsoft CEO Satya Nadella addressed the necessity of overcoming capacity constraints to meet anticipated AI demand, noting, “As AI becomes more efficient and accessible, we will see exponentially more demand.” The company has earmarked $80 billion for AI initiatives in the fiscal year.
DeepSeek’s advancement has led to questions about the effectiveness of US export controls on technology, as the startup achieved its breakthrough using less advanced chips.
Despite DeepSeek’s cost-effective approach, US tech executives maintain that substantial investments are essential to develop scalable AI solutions that meet diverse corporate requirements, Reuters reported. Analysts suggest that while DeepSeek’s model is notable, the extensive infrastructure and resources of companies like Microsoft and Meta position them to deliver comprehensive AI services across various sectors.
Investor reactions have been mixed; Microsoft’s shares declined by 5% following indications that growth in its Azure cloud services may not meet expectations. Meta reported a strong fourth quarter but provided a conservative sales forecast for the upcoming period.
