A single memo turned AI fluency into a hiring constraint. Eight months after Tobi Lutke told Shopify to prove AI can’t do the work before hiring a human, the rest of the industry is adopting the same metric—and not for the reasons the skeptics assumed.
The memo was a filter. By making AI usage a performance metric, Lutke wasn’t trying to make Shopify more efficient—he was reshaping who would want to work there and who would thrive once they arrived. The productivity gains, if they come, are a second-order effect of assembling a team adapted to the new reality.
And now that filter is spreading. Meta. Microsoft. Google. Nvidia. Each week brings another company formalizing what was already happening. Job postings requiring AI skills doubled from 5% to 9% in a single year. Workers in occupations requiring AI fluency grew from one million to seven million. Just this week, Josh Miller at The Browser Company mentioned paying premiums for people who are “native to the Claude Code way of building.”
The evidence on productivity itself? Genuinely mixed. One rigorous study found developers using AI took 19% longer than those working without it. Another found bottom-quartile support reps got a 35% throughput lift while veterans saw almost no gain. AI amplifies variance rather than raising averages—but the hiring market doesn’t optimize for averages. It optimizes for the possibility of outliers.
Here’s what’s inside:
The Red Queen logic. What Lutke actually said in April 2025, and why the memo wasn’t a new philosophy but an existing one applied to a new tool.
How Shopify made it real. The infrastructure—internal LLM proxy, 24+ MCP servers, open-sourced tooling—that made the mandate feasible where most companies would have failed.
The copycat wave. How Duolingo’s version triggered immediate backlash, Fiverr’s maximum-bluntness approach led to 30% layoffs, and Box found a middle path that may prove more durable.
Big tech standardizes the metric. The specific announcements from Meta, Microsoft, Google, and Nvidia—and what Jensen Huang said to managers telling employees to use less AI.
The productivity paradox. Why mixed data doesn’t slow the market—and how variance becomes a hiring premium.
The talent market data. How job postings, skills requirements, and compensation structures are already changing—including the entry-level squeeze intensifying even as companies claim they can’t find AI-fluent talent.
Where this heads in 2026. Role boundaries dissolving, new coordination roles emerging, compensation polarizing, and why the training gap is becoming a strategic liability.
The filter is set. Now we find out who passes.
Listen to this episode with a 7-day free trial
Subscribe to Nate’s Substack to listen to this post and get 7 days of free access to the full post archives.
