On the All-In podcast earlier this month, Salesforce CEO Marc Benioff made one of the most direct statements any major enterprise leader has made about the AI shift in workforce spending: Salesforce expects to spend close to $300 million on Anthropic tokens in 2026, and the company is simultaneously freezing software engineer hiring.

The podcast episode dropped on May 15. The tech industry has been talking about it since.

What Benioff Actually Said

Benioff said the bulk of that $300 million is tied to coding work — AI agents powered by Anthropic’s Claude models completing development tasks that would previously have required additional headcount.

He pointed to productivity gains of over 30 percent across Salesforce’s engineering teams as the reason they stopped hiring engineers in 2025. That freeze is continuing through 2026. With roughly 15,000 engineers currently on staff, those teams now work alongside AI tools including Claude, OpenAI Codex, and Cursor rather than growing in number.

Benioff was careful to note that AI has not replaced engineers. The role has shifted. Engineers at Salesforce are increasingly in oversight and orchestration roles — directing AI agents, reviewing output, and focusing on higher-order problems rather than writing code line by line.

But the budget tells its own story. Three hundred million dollars in AI compute spending where engineering headcount growth used to go.

The Number Behind the Number

Agentforce — Salesforce’s AI agent platform — crossed approximately $800 million in annual recurring revenue. AI now accounts for an estimated 30 to 50 percent of the company’s overall engineering workload.

Salesforce is not scaling back its headcount across the board, either. The company plans to bring on between 1,000 and 2,000 salespeople to support AI product adoption among enterprise customers. What has changed is where the growth investment flows.

Instead of expanding engineering capacity by hiring, Salesforce is buying compute time from an AI provider. The economics make sense to Benioff: AI tokens that do coding work at scale, with existing engineers supervising the output.

This Is Not a Salesforce Story

Salesforce is the clearest, loudest example right now — but the dynamic is showing up across large enterprises. Intuit. Meta. PayPal. Companies announcing layoffs or hiring freezes in technical roles while simultaneously increasing AI spending. The pattern is consistent enough that it should shape how any business leader is thinking about team growth over the next 12 to 24 months.

The question is no longer whether AI will change how businesses staff. It already has. The question is what the right ratio of humans to AI capacity looks like for your specific business.

What This Means for Business

If you run a business of any size, this is worth sitting with.

Salesforce is a company with 15,000 engineers and billions in revenue making a deliberate choice to redirect hiring budget to AI compute. That is not a cost-cutting move — Agentforce is growing fast, and they are investing heavily in the platform. It is a statement about where productivity growth comes from now.

For smaller businesses and mid-market companies, the implication is even more direct. The productivity gains that used to require additional headcount can increasingly come from AI tools, agents, and infrastructure.

This does not mean you replace your team. Benioff explicitly said AI cannot yet function without human oversight. What it means is that the shape of your team can change — and probably should. Humans focusing on judgment, relationships, and oversight. Agents handling execution, research, drafting, and repetitive workflows.

Enterprise DNA works with businesses navigating exactly this transition. Whether you are looking to understand what AI agents could do in your operations, build custom tools on top of AI infrastructure, or upskill your team to work effectively alongside AI, the shift Benioff is describing is real and it is already here.

If you are starting to rethink how your team is structured around AI, there is a clear sequence that matters before you post any AI job listing — one that prevents the mis-hires and half-built deployments that follow when businesses move too fast.

The companies that get this right in 2026 will not be the ones who adopted AI earliest. They will be the ones who figured out the right division of labor between their people and their AI systems.

---

Interested in mapping out what AI could do in your operations? Book a discovery call with the Enterprise DNA team.