Technology thesis · Computing Infrastructure
high conviction matureServerless computing
Serverless has won the developer-experience argument and is now converging with servers via consumption pricing; the 2026 contest is who hosts AI-agent and GPU workloads, not cold starts.
Position maintained continuously · last reviewed Jun 24, 2026
The thesis
Core thesis
Serverless eliminates server management – developers ship functions and the cloud handles scaling and billing. AWS Lambda remains the reference platform. The original event-driven sweet spot still holds, but the historic limitations have eroded: cold starts are largely solved at the edge, durable execution (Lambda Durable Functions) handles stateful multi-step work, and AI inference is now a serverless category in its own right (Modal, Bedrock, Cloudflare Workers AI). The live tensions are cost-at-scale and vendor lock-in, not whether serverless can do the work.
State of the art (2026)
Serverless in 2026 is converging with the servers it once abstracted away. AWS Lambda now ships Durable Functions for stateful multi-step workflows and Managed Instances that pin functions to dedicated EC2 capacity, while Vercel's Fluid compute – billing only active CPU and pausing idle instances – powers over 45 billion weekly requests and cuts costs up to 95%. The frontier has moved from cold-start latency to two questions: who hosts AI-agent and GPU-inference workloads (Cloudflare Workers AI, Modal, Bedrock), and who owns the serverless data layer. Databricks agreed its roughly $1bn Neon acquisition in May 2025, folding serverless Postgres into Lakebase for AI agents. Consumption pricing, not FaaS, is now the defining pattern.
Everything below is live inside CanaryIQ
The full analysis behind the verdict — the structure is real; the content unlocks when you log in.
Signal stack
Evidence stacked leading → lagging
Technology-native KPIs
Metrics that predict trajectory, tracked over time
Landscape map
Who builds what — and who depends on whom
Catalyst calendar
Dated events that will move the position
Technology roadmap
Milestones on the path to maturity
Watchlists
Companies, people and papers — each with a remove-by condition
Decision frameworks
The same call, framed for your desk
Thesis changelog
When our view changed, and why
Change our mind
2 disconfirming conditions
The rest is inside
You've read the verdict. The file is much deeper.
The full signal stack, technology-native KPIs tracked over time, the landscape of who depends on whom, the dated catalyst calendar, decision frameworks for every desk, live watchlists and the changelog of every time our call on Serverless computing has changed — all live inside CanaryIQ.