And marketplaces are how Agentic AI will scale.

Where we are now

Deloitte recently polled two groups: enterprise AI leaders and the broader community.

Both see agentic AI’s promise, but the blockers differ:

1. Nearly 60% of AI leaders point to two practical hurdles— integration into legacy systems and risk/compliance—closely followed by talent gaps.

2. The broader community agrees risk is high but puts “unclear use case/business value” as the number 1 issue.

Translation: leaders are wrestling with plumbing and policy; users are still asking “what’s the first valuable job we can truly automate?”

What’s next

Deloitte projects that about 25% of companies already using GenAI will launch agentic-AI proofs of concept in 2025, trending toward ~50% within two years.

That signals steady expansion—not hype—provided governance and business alignment come first.

But here where it gets interesting

Deloitte argues that the cleanest way to embed and manage agents is through AI agent marketplaces. This can be an internal “app store” that vets, versions, monitors, and governs agents before they go live.

Or this could be hyperscaler agent marketplaces, which Amazon Web Services (AWS), Microsoft Azure, and Google Cloud have now rolled out.

Why marketplaces can scale AI Agents—without the chaos

• Built-in guardrails, not bolt-ons: security, compliance, approvals, version control.

• Clear visibility into who’s using what, where, and how: Built-in monitoring, tracking, and analytics features to provide visibility into AI agent usage across the organization.

• A centralized management layer: A foundation for scaling the agent-driven management, monitoring, and optimization of other AI agents while keeping everything running smoothly.

Deloitte predicts that leading organizations will "simultaneously scale adoption, ensure compliance, and equip people to collaborate effectively with agents."

Practical path

Instead of promising “thousand agents”, perhaps we should start with the promise of a managed environment.

Start with a few high-value agents, prove ROI, and scale through a marketplace.

That’s the path from cautious pilots to durable, revenue-aligned adoption.

And this is where hyperscalers are focused now

As Deloitte notes:

“External marketplaces are starting to emerge as the open, go-to platforms for discovering, publishing, and subscribing to AI agents that can automate workflows, analyze data, and interact with customers….

As these platforms grow, they’ll likely play a big role in how enterprises source and scale AI capabilities—but only if integrated into a secure, governed internal environment.”

What’s your take?

Source: AI Agents Research & Why AI Agents need Marketplaces