AI不是降本工具?英伟达高管一句话戳破行业神话——AI成本可能暴涨至7.9万亿美元,但效率仍未超过人类

‘The cost of compute is far beyond the costs of the employees’: Nvidia executive says right now AI is more expensive than paying human workers 

“计算成本远超员工成本”：英伟达高管称，目前人工智能比雇佣人类员工更昂贵

April 28, 2026 

Recent tech layoffs would initially appear to indicate the great labor shift from human workers to AI may already be happening.

近期科技行业接连裁员，乍看之下，似乎意味着企业正开始把人力资源从员工转向人工智能，一场劳动力结构变革已经启动。

Meta announced last week in a memo that it plans to lay off 10% of its workforce, about 8,000 employees, as well as scrap plans to hire for 6,000 open positions. It’s part of an effort to “run the company more efficiently and to allow us to offset the other investments we’re making,” according to the memo. Microsoft has offered thousands of its own employees a voluntary buyout, the largest the company has ever offered. 

Meta上周在一份内部备忘录中宣布，计划裁减约8000名员工，占总员工数的10%，同时取消6000个空缺岗位的招聘计划。备忘录称，此举旨在“提升公司运营效率，并为其他重点投资腾出资源”。与此同时，微软也向数千名员工提出自愿离职买断方案，规模为公司历史之最。

Other tech headers, however, suggest that right now, AI isn’t saving companies money on labor; it’s actually costing them more than the humans they currently employ. 

然而，科技行业的另一种声音指出，现阶段人工智能并未真正帮助企业降低人力成本，反而可能比继续雇佣员工更加昂贵。

“For my team, the cost of compute is far beyond the costs of the employees,” Bryan Catanzaro, vice president of applied deep learning at Nvidia, recently told Axios. 

英伟达应用深度学习副总裁布莱恩·卡坦扎罗近日在接受Axios采访时表示：“就我的团队而言，计算资源的成本远远高于员工成本。”

An MIT study from 2024 backs up Catanzaro’s experience. Analyzing the technical requirements of AI models needed to perform jobs at a human level, researchers found that AI automation would be economically viable in only 23% of roles where vision is a primary part of the work. In the remaining 77% of the time, it was cheaper for humans to continue their work. 

麻省理工学院2024年的一项研究，也印证了卡坦扎罗的判断。研究人员在评估AI模型达到人类工作水平所需技术条件后发现：在以视觉任务为核心的岗位中，仅有23%的场景具备AI自动化的经济可行性；其余77%的情况，仍由人类完成工作更划算。

In other instances, AI has proved to be fallible, with one engineer saying an AI agent destroyed his database and network as a result of what he called “overuse.” 

此外，人工智能在实际应用中的不稳定性也屡有体现。一名工程师表示，由于其所谓的“过度使用”，某个AI智能体甚至破坏了他的数据库和网络系统。

Despite no clear evidence of AI improving productivity and, according to the Yale Budget Lab, no widespread data to support the idea of AI displacing jobs, Big Tech firms have continued to pour money into AI, announcing $740 billion in capital expenditures this year so far, according to Morgan Stanley, a 69% increase from 2025. The magnitude of spending has caused some companies to rethink their budget altogether. 

耶鲁大学预算实验室也指出，暂无广泛数据支持“AI正在大规模取代岗位”的说法。尽管目前尚无明确证据证明AI显著提升了生产效率，但科技巨头仍在持续加码投入。摩根士丹利数据显示，今年迄今已宣布的AI相关资本支出高达7400亿美元，较2025年增长69%。如此惊人的投入规模，已迫使部分企业重新审视整体预算安排。

“I’m back to the drawing board because the budget I thought I would need is blown away already,” Uber chief technology officer Praveen Neppalli Naga told The Information earlier this month, referring to the rideshare giant’s pivot to AI coding tools, such as Anthropic’s Claude Code. 

优步首席技术官普拉文·内帕利·纳加本月早些时候接受《The Information》采访时表示：“我不得不重新制定预算，因为原本预计的资金需求早已被彻底打破。”他所指的是优步转向使用Anthropic旗下Claude Code等AI编程工具后，成本迅速攀升。

This increase in spending has coincided with more layoffs in the tech sector. According to data from Layoffs.fyi, there have been more than 92,000 layoffs in tech in 2026 so far across nearly 100 companies. The rate of these workforce reductions is already far outpacing that of last year, which saw about 120,000 layoffs in total. 

在支出飙升的同时，科技行业裁员也同步加剧。Layoffs.fyi数据显示，2026年至今，近100家科技公司累计裁员已超过92000人。目前裁员速度已明显超过去年同期，而2025年全年裁员总数约为120000人。

The continued AI spending and layoffs, even as human labor remains cheaper, expose a meaningful discrepancy in the economics of AI, said Keith Lee, an AI and finance professor at the Swiss Institute of Artificial Intelligence’s Gordon School of Business. 

瑞士人工智能研究院戈登商学院人工智能与金融学教授基思·李表示，在人力成本仍具优势的背景下，企业一边加码AI投资、一边持续裁员，暴露出AI商业逻辑中的深层矛盾。

“What we’re seeing is a short-term mismatch,” Lee told Fortune. 

他在接受《财富》杂志采访时表示：“我们正在看到一种短期错配现象。”

The AI-labor cost balance

AI与劳动力的成本平衡

According to Lee, the cost of using AI has remained less efficient than human labor owing to hardware and energy raising operating costs for providers. At its current pace, AI expenditures may reach $5.2 trillion by 2030, with $1.6 trillion from data center spending and $3.3 trillion from IT equipment, according to McKinsey data. Spending could surge to $7.9 trillion by 2030 at an accelerated pace. Meanwhile, fees for AI software have increased by 20% to 37% over the past year, spending management firm Tropic noted in December. 

基思·李指出，由于硬件采购与能源价格持续推高服务商运营成本，AI在成本效率上至今仍难与人类劳动力竞争。麦肯锡数据显示，按当前趋势，到2030年全球AI相关支出可能达到5.2万亿美元，其中数据中心投资约1.6万亿美元，IT设备支出约3.3万亿美元；若发展进一步提速，这一数字甚至可能升至7.9万亿美元。与此同时，企业使用AI软件的费用也在上升。支出管理公司Tropic去年12月指出，过去一年间，AI软件收费上涨了20%至37%。

AI companies may also be losing money as a result of their flat subscription model, Lee noted, with fixed subscription fees failing to cover operating costs for heavy AI users. 

基思·李还表示，许多AI公司采用固定订阅收费模式，可能本身就在亏损。对于高频重度用户而言，统一订阅费往往无法覆盖其实际产生的算力与运营成本。

“As a result, some firms are beginning to reevaluate AI not as a clear cost-saving substitute for labor, but as a complementary tool—at least until the cost structure stabilizes,” he said.

他说：“因此，一些企业开始重新定位AI，不再将其视为明确替代劳动力的降本工具，而是把它当作辅助型生产工具，至少在成本结构趋于稳定之前如此。”

While AI may cost more than human labor today, there will be warning signs of a tipping point toward AI’s economic viability. For one, Lee indicated, the cost of using AI will become significantly lower, with performing inference—how AI analyzes data—for a large language model with 1 trillion parameters plummeting by more than 90% over the next four years, according to a report last month from analyst firm Gartner. 

尽管目前AI成本仍高于人类劳动力，但其迈向经济可行临界点前，可能会出现若干明确信号。基思·李指出，首先是使用成本将显著下降。根据市场研究机构高德纳上月发布的报告，未来四年内，对拥有1万亿参数的大型语言模型进行推理计算的成本，预计将下降90%以上。

AI infrastructure will likely improve, and model designs and hardware supply will follow. AI companies will also likely change how they price their tools, switching from a flat subscription to usage-based pricing, Lee predicted. 

与此同时，AI基础设施将逐步完善，模型架构与硬件供应链也会持续优化。此外，AI企业的收费方式也可能发生变化。基思·李预计，未来行业将逐步从固定订阅制转向按使用量计费，以更准确匹配成本结构。

But the future of AI’s economic viability will also depend on whether the technology proves its worth. It will have to prove itself reliable, with fewer hallucinations and a reduced need for human oversight, effectively integrating into a company’s infrastructure, according to Lee. Federal Reserve data shows about 18% of companies had adopted AI tools as of the end of 2025, a 68% growth in the adoption rate since September 2025.

不过，AI能否真正具备经济可行性，最终仍取决于其能否证明自身价值。基思·李认为，AI必须展现出更高可靠性，减少“幻觉”问题，降低对人工审核的依赖，并能够顺畅融入企业现有系统和业务流程之中。美联储数据显示，截至2025年底，约18%的企业已采用AI工具，较2025年9月增长68%。

“It’s not just about AI becoming cheaper than humans,” Lee said. “It’s about becoming both cheaper and more predictable at scale.”

基思·李总结道：“问题不只是AI何时比人类更便宜，而是它能否在大规模应用下，同时做到更低成本与更高可预测性。”