During discussions with investors, executives from major technology companies expressed their opinions on DeepSeek in various ways. While some praised the artificial intelligence models, others did not think highly of it or avoided elaborating on the topic. Most of them acknowledge that DeepSeek’s affordable models will eventually reduce AI costs, but they seem determined to make massive investments in AI infrastructure by 2025. The timing of when these expenditures will pay off remains uncertain. Microsoft (MSFT) CEO Satya Nadella spoke highly of DeepSeek during a meeting on January 29, while CEO Mark Zuckerberg remained aloof about DeepSeek.

DeepSeek’s innovations caused such a tremor in the US technology sector that the darling of AI, Nvidia (NVDA), lost a $600 billion market value in a single day—the largest daily loss in stock market history. Investors fear that US giants could reduce their AI hardware expenditures. According to the Wall Street Journal, some US lawmakers are now making efforts to ban the use of DeepSeek applications on government devices. Discussions with investors revealed differing opinions among executives of major tech companies regarding DeepSeek. While some lauded AI models, others seemed indifferent or avoided details. Most agree that the cheap models of DeepSeek will eventually lower AI costs, but they appear committed to making massive investments in AI infrastructure by 2025. The uncertainty surrounding when these investments will pay off continues. Top technology company executives’ statements on the matter:

Google (GOOG) CEO Sundar Pichai, when asked about the impact of DeepSeek’s low-cost model on Google, stated: ● The DeepSeek team has done “a commendable job”. ● However, he indicated that Google’s own Gemini Flash models are more efficient. ● He highlighted Google’s strength in end-to-end optimization, stating, “Our 2.0 Flash models are extremely efficient compared to DeepSeek’s V3 and R1 versions.” ● He expressed anticipation for the increase in usage scenarios and opportunities with the lowering of AI usage costs.

Amazon (6 February) CEO Andy Jassy disagreed with the belief that AI spending will decrease with DeepSeek’s new models: ● He mentioned that the assumption that “technology expenditures decrease when the cost of technology components decreases” has never been historically verified. ● Using aggressive cloud investments from the early 2000s as an example, he mentioned that despite cost reductions, expenditures increased instead of decreasing. Indeed, executives of Meta, Alphabet, Amazon, and Microsoft announced that although training and inference costs of AI models are expected to decrease, they plan to spend a total of $325 billion by 2025, indicating a 46% increase from the previous year. Among them, Amazon is set to spend the most.