# Senior Engineers Are the Next DRAM Shortage > AI is eating junior engineering roles, but senior engineers are the next scarce resource, build your own to avoid price spikes. - Title: Senior Engineers Are the Next DRAM Shortage · Greg Herlein - Summary: AI is eating junior engineering roles, but senior engineers are the next scarce resource, build your own to avoid price spikes. Senior engineers are the next… - Keywords: senior engineers, agentic coding, mentorship, software engineering, career, leadership, training, judgment, productivity, chaos drills, ckas, Senior - Source: Greg Herlein — https://blog.herlein.com/post/build-another-engineer - Published: 2026-09-18T08:00:01+00:00 - Read time: 2 min - Topics: senior engineers, ai, agentic coding, mentorship, software engineering, career, leadership, training, judgment, productivity, chaos drills, ckas ## Senior Engineers Are Next AI is eating junior engineering work, but senior engineers are the next scarce resource. Demand for judgment-driven agents will explode while supply remains flat. > Senior engineers are next, and the only hedge is to build your own, starting now. ## The Junior Engineer Bundle Hiring juniors historically provided two things: the work they did and the senior they became through on-the-job training. > The work they did. Real tickets, real fixes, real features. Not glamorous, but it shipped, and it roughly paid for their salary. ## AI Splits the Bundle AI automates the junior work, making it redundant and silently killing the on-the-job training pipeline that builds seniors. > AI made the junior's labor redundant, and companies are about to throw out the junior's training value along with it. ## The DRAM Analogy DRAM prices rose 400% from 2024 to 2026 due to a shortage that took years to fix, senior engineers face the same crisis. > DRAM prices rose more than 400% from the start of 2024 through the end of 2026, nearly doubling in a single quarter. ## The Accelerant Senior engineers paired with nine agents create a 3x productivity boost, but this model burns through the senior's time without growing new talent. > One senior, nine agents, 3x. If you have five seniors, you have five of those. That's linear. ## The Trap The nine-agent model is linear and self-sustaining, it doesn't grow the senior pipeline, so the shortage worsens. > The nine-agent senior isn't just linear. Left alone, it's decreasing, a fab quietly running its own supply down. ## The Hedge The only hedge is to build your own senior engineers through mentorship, creating a branching pipeline that compounds over time. > One senior who mentors three juniors, each driving their own agents, creates a curve that compounds over time. ## Real Work for Juniors Juniors should build small, contained, testable work that matters to the business, not toys or abstract concepts. > Canary tests, smaller services, SDKs, example clients that double as test harnesses. ## Chaos Drills Juniors need hands-on failure drills to build judgment, deliberately breaking systems and observing outcomes under time pressure. > Chaos drills: kill a node mid-rollout, partition the network, and so on, on a cluster safe to wreck. ## Certs as Forcing Functions Use certifications like CKA and AWS SAA as live, time-bound challenges to prove judgment under failure, not just as badges. > Certs as a forcing function: a performance-based exam puts a human alone at a live terminal with a broken system and a clock. ## No-AI Checkpoints At least one exercise per skill area with AI turned off to verify independent mental models without tool dependency. > No-AI-assist checkpoints: at least one exercise per skill area with agents turned off, under time pressure. ## AI as Tutor Use AI to learn 10x faster, but only to build judgment, not to replace the mental model through direct knowledge. > Stuck on how Raft handles a split vote? Ask. Can't see why the NetworkPolicy is silently dropping traffic? Have the model walk you through it. ## The Rickover Legacy The US Navy's Rickover program trained junior engineers to handle nuclear reactors through failure drills, a model that applies to modern software. > The Rickover program didn't work because the kids memorized procedures. It worked because it was built around rising standards, constant training, and escalating bad news fast. ## Why Seniors Hate It Letting juniors build agentically is slower than doing it yourself, but it creates long-term capacity that senior engineers need to maintain. > You will watch them go down a road you'd have skipped. You will want to grab the keyboard. ## The Trade-Off Short-term velocity for long-term capacity: mentoring pays a tax now but compounds later, creating a sustainable pipeline. > The mentor pays a tax now and compounds later. You can't learn to swim from a book. ## The Real Answer The highest-leverage action in the agentic era is building another engineer through mentorship, not just building faster. > The highest-leverage thing you can do in the agentic era is still, stubbornly, to build another Engineer, not just build it yourself faster. ## Key takeaway Senior engineers are the next DRAM shortage: demand for judgment-driven agents will explode while supply remains flat, so companies must build their own to hedge.