charles.lin
531 posts

charles.lin
@linxiaocong
前QQ后台研发,腾讯云IM研发。现在在香港躺平。

又要失业了,真离谱。公司大力推了几个月 AI 后,开始图穷匕见了。早知道有这一天,没想到来的这么快。🥲 我一直觉得这种状态是不可持续的,平白无故的每个月给每个人几百刀的 Token,这个成本是实打实的支出了的,工作效率也实打实的提升了不知道多少倍的。想要维持旧的人员规模是不可能的。



The layoff wave tells two stories, not one. Tech giants like Amazon, Meta, and Microsoft are cutting to fund GPU purchases. Their revenues are growing. Their stock prices are climbing. They're firing people to free up cash for compute. This isn't cost-cutting during a downturn. It's a forced reallocation from payroll to datacenter capacity. The math is brutal: every percentage point of headcount reduction funds another batch of H100s. Meanwhile, UPS, Nestle, Ford, and Target are cutting for the opposite reason. They've already deployed AI tools that work. Customer service automation, supply chain optimization, generative design systems. The productivity gains are real and compounding. These companies don't need to buy massive GPU clusters. They're renting inference from hyperscalers and cutting headcount because the math finally works. Both sides are feeding the same beast. Tech companies are buying the shovels. Everyone else is buying the gold those shovels dig up. Semiconductor companies sit in the middle, collecting rent from the entire value chain. TSMC, NVIDIA, and ASML are printing money while employment craters on both ends. The timing matters. We're at 10% enterprise AI adoption, heading toward 50%. History says this phase moves fastest and generates the most wealth. But that wealth is concentrating in compute, not labor. The gap between market cap growth and wage growth has never been wider. This isn't a recession. It's a rebalancing. And most workers are on the wrong side of it.

















