The venture capital flywheel just reversed direction, and customer support automation is apparently worth more as an acquisition target than as an independent unicorn.
The Summary
- Decagon, valued at $4.5 billion, hired an investment bank to field acquisition interest from ServiceNow and other buyers, signaling a new phase where established tech giants buy AI capabilities instead of waiting to build them
- The company could pivot to fundraising if M&A talks fall through, but recent deals like Nvidia acquiring Hugging Face for $12.9B and Stripe buying OpenRouter for $8B show AI companies are commanding premiums well above their last valuations
- This marks the maturation of the AI agent market: when enterprise software incumbents start writing checks for customer support automation, the technology has crossed from experimental to essential
The Signal
Decagon builds AI agents that handle customer support for enterprise customers. Not chatbots that escalate to humans after three failed attempts. Actual agents that resolve tickets end-to-end. The company reached a $4.5 billion valuation earlier this year, and now ServiceNow wants in.
ServiceNow makes perfect strategic sense as a buyer. They sell workflow automation to enterprises. Customer support is workflow. Add Decagon's agents to ServiceNow's platform, and suddenly every ServiceNow customer gets AI that actually works integrated into systems they already use. That's not disruption, that's distribution at scale.
"Growing AI-related businesses are often getting offers at a significant premium to their last valuation."
The broader pattern matters more than this single deal. The AI market just shifted gears. For three years, the playbook was: raise huge rounds, scale fast, worry about exits later. Now exits are arriving early, and they're arriving rich. Consider the recent comps:
- Nvidia paid $12.9B for Hugging Face
- Stripe paid ~$8B for OpenRouter
- Both deals came at premiums to prior valuations
These aren't acqui-hires. These are strategic acquisitions by companies that looked at the build-versus-buy calculus and decided building would take too long. The AI agent capability gap between startups and incumbents is wide enough that writing a check makes more sense than spinning up an internal team.
Decagon's position is particularly interesting because customer support is simultaneously high-value and high-pain. Every enterprise already spends millions on support infrastructure. An AI agent that cuts headcount by 40% while improving response times pays for itself in quarters, not years. That math works at $4.5 billion. It probably works at $6 billion.
The calculus for Decagon's founders is straightforward: sell now at a premium, or raise another round and bet on 3-5 more years of execution risk. In a market where Hugging Face and OpenRouter both sold, waiting starts to look like ego, not strategy.
The Implication
If you're building AI agent infrastructure, you now have a clear exit path. The question isn't whether big tech will acquire, but when and at what multiple. Companies like ServiceNow, Salesforce, and Oracle can't afford to miss the agent transition. They'll pay to catch up.
For founders: hire the bank, take the meetings, understand your leverage. The window where premiums are this high won't stay open forever. For engineers at incumbents: your company is probably already talking to one of these startups. The internal build you've been working on might become the integration layer for someone else's acquisition.