Summary of Key Points
This article directly challenges the prevailing "default rule" in the AI industry: Over the past two years, the entire sector has followed a strategy of "embracing AI while simultaneously cutting staff to reduce costs." Giants like Meta, Microsoft, and WPP have invested billions in AI research and development while significantly reducing their workforce, viewing "using fewer people to do more work" as the core value of AI. However, two companies have defied this common sense by taking the opposite approach: DeepSeek, which focuses on building large-scale AI infrastructure, has just announced a doubling of its team size and is hiring 150 new positions in areas such as backend development and flexible computing for AI agents, without hiring a single AI algorithm researcher; Notion, a note-taking tool company, plans to increase its staff by 30% this year, reaching a total of 1,300 employees—three times the number of employees at some of the largest domestic AI companies. These two companies' unconventional decisions have shattered the stereotype that AI efficiency always leads to smaller companies and that a few people can easily profit from managing AI systems.
---
Simplified Explanation by Dimension
1. Understanding the Context: Why the Industry is Cutting Staff with AI, and Why These Two Companies are Hiring
The standard practice in internet companies has become a fixed routine: managers announce a full embrace of AI during meetings, then assign layoffs to HR under the guise of "optimization"—using AI to replace three people’s work with just two, effectively cutting labor costs in half. For example, Zuckerberg once planned to cut up to 60% of his team, and WPP, the world’s largest advertising agency, laid off 1,000 people to invest the savings in AI. The prevailing logic is that AI’s role is to replace human labor, and the money spent on AI can come from the reduced workforce. However, DeepSeek and Notion are breaking this pattern. They are not using AI to replace existing jobs but to create entirely new businesses that were previously impossible with traditional methods. These new businesses lack existing systems and processes, so they need to be built from scratch by new employees, which cannot be achieved by cutting staff.
2. DeepSeek Hires More Backend Engineers than AI Scientists: The Challenge of Infrastructure is Ten Times Greater than Algorithm Development
Many people think of large AI companies as hiring lots of AI scientists to make AI more intelligent. But DeepSeek’s 150 new hires are all in areas such as backend development, server scheduling, and system optimization. The reason is simple: DeepSeek’s AI algorithms are already established, and the current challenge is not making AI smarter but ensuring that millions of users can use it smoothly without lagging, and that tens of thousands of AI agents can run stably on servers. For example, if you open a food delivery business and develop a popular product, you don’t need to hire more chefs when you suddenly receive 100,000 orders; instead, you need to hire staff to optimize the delivery process, set up a scheduling system, and manage the inventory. DeepSeek is doing exactly that: while small teams could handle a few hundred orders with a few servers, now it needs to support millions of users and tens of thousands of AI agents, requiring a complete overhaul of the infrastructure, which is ten times more work than developing algorithms.
3. Why the Small, Popular Note-Taking Tool Notion Has Three Times More Employees than Large AI Companies
Notion is often seen as a simple, free note-taking tool created by a small team. However, it now has 1,000 employees and plans to grow to 1,300, three times the number of employees at some large AI companies. Notion’s success relied on free usage, with users gradually adopting the tool and then purchasing the corporate version, resulting in a 90% gross margin and steady cash flow. But with the integration of AI, its profit model has changed: each time users use AI to summarize documents or organize tasks, Notion has to pay fees to OpenAI and Anthropic. This has reduced its gross margin by 10 percentage points. Additionally, integrating AI into corporate workflows is complex, requiring sales and support staff to help companies import documents, set up security measures, and customize AI functions to their needs, significantly increasing labor demands.
4. Debunking the Myth: AI Efficiency Does Not Mean Smaller Companies
There’s a widespread belief that AI will reduce the need for employees, leading to smaller companies. These two companies’ hiring decisions refute this: while AI can reduce the cost of tasks by a factor of ten, it doesn’t mean fewer people are needed. Instead, it allows companies to tackle new, previously unfeasible tasks, increasing the total workload. New challenges arise, such as managing millions of AI requests, ensuring data security, and providing training for users. These new requirements create new job opportunities. In other words, AI frees up existing staff but opens up new markets with unsolved problems, necessitating more personnel to handle them.