---
title: "OpenAI and Anthropic are hiring Salesforce's sellers"
date: 2026-06-18T09:00:00+02:00
language: en
slug: 2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia
url: https://mathieuhaye.fr/blog/en/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia
alternate: https://mathieuhaye.fr/blog/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia
category: Business & Growth
description: "OpenAI and Anthropic have hired nearly 100 Salesforce staff in 18 months, mostly in sales. What this talent raid reveals about the real AI enterprise race."
---

# OpenAI and Anthropic are hiring Salesforce's sellers

> OpenAI and Anthropic have hired nearly 100 Salesforce staff in 18 months, mostly in sales. What this talent raid reveals about the real AI enterprise race.

According to a report from [The Information](https://www.theinformation.com/articles/openai-anthropic-tap-salesforce-talent) published on June 17, 2026, OpenAI and Anthropic have hired nearly 100 Salesforce employees over eighteen months, mostly in sales and go-to-market, a sign that the AI battle is shifting from the model to enterprise distribution.

For two years the dominant story said the AI war would be settled on parameters, benchmarks and cluster size. The hiring wave revealed this week tells a different story: the two best-capitalised labs on the market are draining a software vendor's commercial ranks, not its research teams.



## What happened



According to [The Information](https://www.theinformation.com/articles/openai-anthropic-tap-salesforce-talent), relayed by [Benzinga](https://www.benzinga.com/markets/private-markets/26/06/53261205/openai-anthropic-poach-approximately-100-salesforce-employees-this-year) on June 17, 2026, more than 45 former Salesforce staff have joined Anthropic since the start of 2026, and close to 40 have joined OpenAI in the same window. The total reaches roughly 100 departures over eighteen months. Almost all the roles sit in sales, marketing, go-to-market and revenue operations. Per [Crypto Briefing](https://cryptobriefing.com/openai-anthropic-hire-salesforce-employees/), commercial functions account for about 20% of open positions at both labs.

The names involved are not minor. Denise Dresser, former CEO of Slack under Salesforce, became Chief Revenue Officer at OpenAI. Brian Landsman, who ran AgentExchange at Salesforce, moved to OpenAI as Vice President of Global Partnerships. Jason Boehmig, co-founder of Ironclad, now leads the legal-industry product team at OpenAI. These are not interns; they are leaders who know how large enterprises actually buy.

The stated reason is simple: these profiles bring **established relationships with Fortune 500 companies**. While its sellers leave, Salesforce has, according to these sources, slowed some engineering hiring. And CEO Marc Benioff said he would probably spend $300m on Anthropic tokens in 2026. So Salesforce is losing salespeople to a vendor it is becoming a customer of.



## Why are AI labs hiring sellers, not researchers?



Because the bottleneck has moved. When GPT-4 and Claude 3 shipped, the competitive edge was raw model quality. In 2026 the frontier models sit within a few points of one another on most business tasks, and a CIO does not sign a seven-figure contract on the strength of an MMLU score. They sign when someone can answer three questions: what does this really cost at scale, where does my data go, and what return can I show my board.

Those questions are not solved with a better model. They are solved by people who have already negotiated multi-million-dollar rollouts, who know the procurement, legal and security teams of large accounts, and who can turn an impressive demo into a multi-year contract. Salesforce spent twenty-five years training exactly this profile. It is the densest pool of enterprise software sellers in the world; it made sense for the labs to draw from it.

Crypto Briefing's observation captures the turn: OpenAI and Anthropic are no longer just building models, they are building the commercial machinery to sell those models directly to large organisations. OpenAI also plans to nearly double its headcount, from about 4,500 people. You do not double a research lab by hiring 4,500 researchers; you do it by building a sales force.



## What Benioff's $300m bet reveals



Marc Benioff's line is the most instructive detail in the story. Salesforce is losing dozens of salespeople to Anthropic, and at the same time its boss announces $300m of spending with that very Anthropic. The two facts do not contradict each other; they describe the new power structure of enterprise software.

The application vendor needs the foundation model to stay relevant, and the foundation model needs the vendor's sellers to reach buyers. Each supplies the other with the missing piece. But the margin is shifting: when a vendor hands $300m a year to its model supplier while that supplier hires away its best sellers, the question of who captures value over time becomes open.

For company leaders, the message is concrete. The criterion for choosing an AI vendor in 2026 is no longer only "which model performs best", but "who can deploy in my environment, handle my privacy and prove the return". The labs understood this before many of their customers did: they are paying for that competence, not for half a benchmark point.



## Distribution, the new front in the AI war



This sequence closes a cycle. The 2023-2025 phase was about capability: who has the biggest model, the longest context, the best reasoning. The phase now opening is about distribution: who can get those capabilities adopted in organisations that have processes, regulatory constraints and a legitimate wariness. Commercial talent becomes a scarce resource, just as chips or researchers were yesterday.

It is also a warning to SaaS vendors. If the value layer moves toward the model and toward the direct customer relationship, the application middleman has to prove it brings more than a wrapper. Salesforce has an answer: customer data, workflows and accumulated trust. But the fact that its best sellers see more upside at a model vendor than inside the incumbent platform is worth noting.



## How this connects to my freelance work



I see the same mechanism at the scale of SMEs and nonprofits, in miniature. When I rolled out Salesforce for 3018, the e-Enfance helpline, with the 3CX telephony integration in Apex and LWC, the AI model was never the hard part. The hard part was adoption: building trust, fitting the tool into existing day-to-day work, proving it holds in production. Same logic on my bilingual Pipedrive mission at Horus, or on the n8n automation that runs market monitoring for Fromagerie Ermitage: the tech is settled in days, trust and usage are built over weeks.

What the labs are paying a premium for is exactly this skill of putting things into production and reassuring buyers. For a client, the lesson is reassuring: a good seller or a good integrator who can answer questions about cost, data and return is worth as much as a frontier model. My job is precisely to hold that bridge between an AI capability and an organisation that has to adopt it without breaking its processes.



## The essentials in 30 seconds



                - OpenAI and Anthropic have hired nearly 100 Salesforce employees in eighteen months, mostly in sales and go-to-market (The Information, June 17, 2026).

                - More than 45 departures to Anthropic and close to 40 to OpenAI since early 2026; commercial functions are about 20% of open roles.

                - Salesforce plans to spend $300m on Anthropic tokens in 2026, per Marc Benioff, while losing its sellers to the same vendor.

                - The AI bottleneck is no longer the model but distribution: selling to large accounts and reassuring them on price, data and return.





**Takeaway.** When the two best-funded AI labs in the world spend their human capital hiring sellers rather than researchers, they are telling us where the value now sits. The next battle will not be about the best model, but about who can get it inside a real organisation.



## Frequently asked questions



### How many Salesforce employees have OpenAI and Anthropic hired?



According to The Information, the two labs have hired nearly 100 Salesforce staff over eighteen months. More than 45 joined Anthropic since early 2026 and close to 40 joined OpenAI in the same window, mostly in sales and go-to-market roles.



### Why are AI labs hiring sellers instead of researchers?



Because the bottleneck is no longer model quality but the ability to sell to large accounts. Salesforce sellers bring established relationships with Fortune 500 companies and know how to answer questions about pricing, privacy and return on investment.



### Why is Salesforce spending $300m on Anthropic?



Marc Benioff said Salesforce would spend roughly $300m on Anthropic tokens in 2026. Salesforce embeds Claude in its products while losing salespeople to the same vendor, which shows how dependent software firms have become on foundation models.

---

Source: [https://mathieuhaye.fr/blog/en/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia](https://mathieuhaye.fr/blog/en/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia) | Other language: [https://mathieuhaye.fr/blog/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia](https://mathieuhaye.fr/blog/2026-06-18-openai-anthropic-debauchent-salesforce-vente-ia)
