The $100B Bet on Agent Infrastructure

$100 billion. That is Bain & Company’s estimate for the US SaaS market specifically driven by agentic AI. This isn’t a warm fuzzies market. This is automating the coordination work that currently runs on people and spreadsheets. Enterprise middleware. The dirty work.

The Infrastructure Play

You don’t publish a $100B estimate and stay on the sidelines. The capital formation is happening at a scale we haven’t seen since the early cloud days. OpenAI just created the OpenAI Deployment Company, raising $4bn from a syndicate including TPG, Advent, Bain Capital, and Brookfield. Nineteen firms. This isn’t venture funding. It is infrastructure financing. Data center megaprojects. Power purchase agreements. The asset managers who buy airports are now buying the runways for AI agents. The money doesn’t lie. It sees the $100B exit ramp.

The Trust Infrastructure

Capital is just fuel. The engine is trust. And the most interesting experiment in agent trust is happening inside Shopify.

Tobias Lütke revealed the inner workings of Shopify’s internal coding agent, River. River has a single hard rule: it never operates in private.

River does not respond to direct messages. She politely declines and suggests to create a public channel for you and her to start working in.

This is the killer feature. Not the ability to write code. Not the speed. The auditability. Every single action River takes is recorded in a searchable, public Slack channel. Every prompt. Every output. Every mistake.

Enterprise agents cannot be black boxes. The people approving vendor security questionnaires will never allow it. The only way an agent scales inside an organization is if its behavior is boringly transparent. Shopify gets this. The rest of the industry is still building magic tricks.

We are also seeing this tension play out in hardware. Record Mac mini demand is being driven by developers running Claude agents locally. The hardware shortage tells us a simple truth: people want control. They want the agent to be local, private, and fully owned. River gives them that trust in the cloud. The Mac mini gives it to them on the desk.

The Signal

The signal for builders is loud and clear. The $100B market doesn’t belong to the best models. It belongs to the best deployment infrastructure.

Can you deploy an agent that can be fully audited? Can your agent operate in public channels without leaking context? Are you building a tool for a single user or a system for an entire organization? The money (OpenAI/TPG) is betting on massive, centralized infrastructure. The successful practitioners (Shopify) are betting on radical transparency. The market size (Bain) justifies both bets.

Stop optimizing for the demo. Start optimizing for the deployment.

The agent era isn’t about building smarter models. It’s about building responsible infrastructure. The $100B goes to the plumbers, not the poets.