由 jpichiban Sat, 01 Feb 2025 13:00:01 GMT
Dado Ruvic | ReutersMicrosoft CEO Satya Nadella commented on its “real innovations.” OpenAI’s Sam Altman described it as “clearly a great model.” Apple CEO Tim Cook said “innovation that drives efficiency is a good thing.” And Palantir’s Alex Karp said it shows the importance of “an all-country effort.”The tech CEOs were all talking about China’s DeepSeek, which burst out of obscurity and into the center of the tech universe this week. In the past few days, those execs and many of their peers have addressed questions about the startup lab’s new artificial intelligence model, which has stunned experts and was reportedly much more cost effective to create than competitive models in the U.S.DeepSeek’s mobile app shot up to the top of the charts on Apple’s App Store early in the week and remained in the lead spot as of Friday, ahead of OpenAI’s ChatGPT. Reports that its new R1 model, which rivals OpenAI’s o1, cost just $6 million to create sent shares of chipmakers Nvidia and Broadcom down 17% on Monday, wiping out a combined $800 billion in market cap.The timing was stark. DeepSeek’s rollout landed just as tech earnings season was about to begin, with Meta, Microsoft, Tesla and Apple all reporting between Wednesday and Thursday, and a week into President Donald Trump’s second term in office. Trump has emphasized the importance of the U.S. winning in AI, particularly against China, and in his first week back in the White House announced a project called Stargate that calls on OpenAI, Oracle and SoftBank to invest billions dollars to boost domestic AI infrastructure.Read more CNBC reporting on AIPerplexity AI revises Tiktok merger proposal that could give the U.S. government a 50% stakeTrump tariffs could raise prices on technology like laptops, smartphones and AIOpenAI introduces Operator to automate tasks such as vacation planning, restaurant reservationsZuckerberg sets Meta’s AI targets for the year, expects to spend $60 billion on growthScale AI CEO says China has quickly caught the U.S. with the DeepSeek open-source modelThere’s been plenty of debate online about the significance of DeepSeek’s rollout and whether the financial achievement is real. A new report from research firm SemiAnalysis estimated DeepSeek’s costs at “well higher than $500M over the company history.”But it was an inescapable topic this week across the industry. AI is every company’s focus right now, particularly in technology, where industry leaders are spending tens of billions of dollars building out data centers and buying advanced chips to develop more powerful models. For leaders of U.S. tech companies, the unifying aspect is the need to beat chief adversary China in the defining technology of the future.”Technology is not inherently good,” and could pose threats in the wrong hands, Karp, the CEO of Palantir, told CNBC’s Sara Eisen in an interview that aired Friday. “We have to acknowledge that, but that also just means we have to run harder, run faster, have an all-country effort.”watch nowVIDEO18:2018:20Watch CNBC’s full interview with Palantir CEO Alex KarpMoney MoversCook was asked by an analyst on Apple’s earnings call if the DeepSeek developments had changed his views on the company’s margins and the potential for computing costs to come down. Cook, whose company had just reported a record gross margin, offered a vague response.”In general, I think innovation that drives efficiency is a good thing,” he said. “And, you know, that’s what you see in that model. Our tight integration of silicon and software, I think, will continue to serve us very well.”A day earlier, Meta CEO Mark Zuckerberg suggested that the overall situation is nuanced and that early reports and results from a single model don’t fundamentally change the equation. Meta said last week that it would invest between $60 billion and $65 billion in 2025 to expand its computing infrastructure related to artificial intelligence.”It’s probably too early to really have a strong opinion on 扎克伯格在公司第四季度财报电话会议上谈到DeepSeek时表示：“这意味着基础设施和资本支出的轨迹将如何发展，这里有一系列趋势正在同时发生。”
在微软的电话会议上，纳德拉在开场白中表示，想要运行DeepSeek R1的人“很快就能”在公司的Copilot+ PC和Windows的图形处理单元（GPU）上实现。随后，一位分析师询问纳德拉，我们是否“正在看到AI以更低的成本扩展”。
纳德拉回应说：“AI正在发生的情况与常规计算周期没有什么不同，”并补充说，“这总是关于弯曲曲线，然后在曲线上增加更多的点。”他说DeepSeek展示了一些“真正的创新”，而微软支持的OpenAI也看到了类似的改进。最终，他说，“这一切都会商品化”，客户是主要受益者。
LinkedIn联合创始人、OpenAI的早期投资者、微软董事会成员、Inflection AI的联合创始人里德·霍夫曼告诉CNBC，现在不是恐慌的时候。霍夫曼本周推出了他的最新AI初创公司Manas AI，获得了近2500万美元的支持，其使命是尝试加速药物发现过程。
霍夫曼表示，虽然DeepSeek可能会鼓励美国公司加快步伐并更早分享他们的计划，但新的发现并不表明大型模型是一个糟糕的投资。“竞争游戏开始了，”他说。“但我不认为这是‘哦，我的天，我们输了！’的美国技术。”
OpenAI可能是最直接的竞争对手，首席执行官阿尔特曼在周四在华盛顿特区的一个活动中称R1“显然是一个伟大的模型”。他还回应了特朗普总统表达的情绪，特朗普表示DeepSeek应该是美国科技的“警钟”。
“这是一个提醒，提醒我们竞争的水平和民主AI需要获胜，”阿尔特曼说。“这是一个强大的模型，我认为这也是对推理兴趣水平和开源兴趣水平的提醒。”
本周早些时候，阿尔特曼在X上发文，强调OpenAI将继续推进的意图。“我们显然会提供更好的模型，”他在1月27日的帖子中写道。“而且有一个新的竞争对手真的很激励人心！”
— CNBC的乔丹·诺维特、海登·菲尔德、基夫·莱斯温、阿什利·卡普特和乔纳森·瓦尼安为本报告做出了贡献。观看：Nvidia的未来聚焦。
原文链接：https://www.cnbc.com/2025/02/01/deepseek-shocked-ai-world-this-week-heres-how-tech-ceos-responded.html