Our Take
Garman's argument that AI will change jobs, not eliminate them, is undermined by Amazon's own roadmap: 30,000 corporate layoffs since October, plans to replace half a million jobs with robots, and CEO Andy Jassy's explicit statement that AI will 'reduce our total corporate workforce.'
Why it matters
Enterprise customers are beginning to see real ROI on AI deployments, not just proof-of-concept wins. That transition speed matters because AWS sits beneath most of that infrastructure spend, and the gap between Garman's public optimism and Amazon's cost-reduction strategy signals how management actually views labor displacement.
Do this week
Enterprise leaders: map your next 12 months of operational automation against your hiring plans, then ask your board whether that ratio aligns with AWS or Garman's public statements.
AWS CEO doubles down on hiring junior staff while selling agents that do their work
Matt Garman, CEO of AWS since June 2024, publicly argues that junior employees remain as necessary as ever in the AI economy. Amazon is hiring 11,000 interns and new graduates this year. Garman has called replacing junior staff with AI "one of the dumbest things I've ever heard," and told Wired that avoiding junior hires is a "non-starter for anyone trying to build a long-term company."
He frames AI the way he frames cloud adoption: a technology that eliminates specific jobs but ultimately expands the labor force. Excel eliminated hand calculators but created new roles for people who learned the tool. Jobs change, he argues, but don't vanish.
The tension is stark. AWS now sells software agents that recruit, code, process claims, and optimize workflows. Amazon itself has cut roughly 30,000 corporate jobs since October 2023. CEO Andy Jassy has written that AI will "reduce our total corporate workforce" in the years ahead. The company plans to replace half a million jobs with robots.
Customer ROI is finally visible, which changes the timeline
Most enterprises spent two years running proof-of-concepts with minimal business returns because they had no rollout plan. That phase is ending. At a recent roundtable of roughly 100 CIOs, Garman reported that 90 percent said they either see materially positive ROI on AI investments today or have a clear path to "really high ROI" in the next couple of months (per his account in the interview). A year prior, nearly all viewed AI as a pure cost.
Coding and software development show the clearest wins. But the real shift is happening one step up: agents that use code-writing ability to autonomously handle business processes. Garman cited telco network optimization and loan processing workflows that have moved from 20% success rates to 80% to high-90s. Companies are now extending these agents into adjacent workflows.
That speed matters. Garman notes that cloud adoption took two decades to displace on-premises workloads. He expects AI to move much faster because cloud infrastructure is already in place and most data lives in it. The compounding effect: cloud + AI adoption feedback loop accelerates the job-displacement timeline, even if new roles emerge.
Map your automation roadmap against your labor plan
Garman's optimism rests on an untested assumption: that new jobs appear as fast as old ones vanish, or that displaced workers retrain quickly enough. Amazon's own hiring (11,000 juniors) and firing (30,000 corporate staff since October) suggest the company is hedging. The 11,000 represents roughly one-third of the 30,000 cuts, not a replacement.
For practitioners: if your enterprise is seeing 90% ROI on pilot agents and you're planning to scale those agents across functions, you need a labor-transition plan, not just a technology plan. Garman's claim that "the most durable skill is a willingness to learn" assumes retraining happens at organizational pace, not AI adoption pace. AWS is building the agents faster than most enterprises can rebuild their workforce.
The real question: is Garman's hiring of 11,000 juniors a bet on new job creation, or a signal that AWS itself knows those junior roles will be different in shape and scale than they are today?