Stick to the Facts
Add Nbsla.ca as a Preferred Source on Google to see more of our stories in your search results.
The AI race inside Silicon Valley just entered another massive phase after Marc Benioff revealed that Salesforce is expected to spend nearly $300 million on tokens from Anthropic this year alone. The statement from Salesforce CEO Marc Benioff is now fueling major discussion across the enterprise AI industry, especially around AI coding agents, AI software development, and the growing dominance of Anthropic’s Claude models in enterprise environments.
Speaking on the All-In Podcast, Marc Benioff made it clear that Salesforce sees AI coding as the future of software creation. According to Benioff, the majority of the company’s enormous spending on Anthropic tokens is tied directly to coding agents and AI-powered engineering workflows.
“These coding agents are awesome. Anthropic is awesome,” Benioff said during the podcast interview. The Salesforce CEO added that software production costs are rapidly falling because AI systems can now generate, improve, test, and ship code at unprecedented speed.
The comments from Marc Benioff immediately attracted attention because Salesforce is one of the world’s biggest enterprise software companies. When a company of Salesforce’s scale commits $300 million in AI token spending, it signals how aggressively major corporations are investing in generative AI infrastructure.
Marc Benioff Says AI Coding Agents Are Changing Everything
According to Marc Benioff, AI coding agents are no longer experimental tools. Instead, they are becoming core systems inside enterprise operations. Salesforce CEO Marc Benioff explained that engineering teams are now moving faster than ever because AI agents can assist developers with writing code, debugging software, testing applications, and accelerating deployment cycles.
Benioff described the productivity gains from AI agents as “unprecedented.” He specifically pointed to engineering, customer service, support, marketing, and distribution as departments already experiencing dramatic operational improvements because of artificial intelligence.
The Salesforce CEO explained that the company can now build, ship, and sell software simultaneously. Before the rise of AI coding agents, that type of rapid iteration was considered impossible inside large enterprise software companies.
The growing use of Anthropic Claude models inside Salesforce shows how enterprise AI adoption is increasingly shifting toward practical productivity use cases rather than experimental chatbot demonstrations. Coding has become one of the biggest drivers of enterprise AI spending because software development generates enormous token consumption.
Why Salesforce Is Spending $300 Million on Anthropic Tokens
Enterprise AI companies charge businesses based on token usage. Tokens are essentially the chunks of text that AI models process while reading prompts and generating responses. Every coding request, software instruction, automation workflow, or AI-generated output consumes tokens.
For a massive organization like Salesforce, token usage can scale extremely quickly when thousands of employees use AI coding agents daily.
Marc Benioff explained that coding is currently the primary use case behind Salesforce’s projected $300 million Anthropic spending. AI-assisted engineering workflows require huge volumes of prompts, context windows, code analysis, and generated outputs, all of which increase token consumption.
The statement also highlights how expensive frontier AI models remain for enterprise customers. Companies are increasingly spending hundreds of millions of dollars annually to access advanced AI systems capable of performing complex reasoning and coding tasks.
Anthropic’s Claude models have gained strong traction among developers because many engineers believe Claude performs exceptionally well on long-context reasoning, code generation, and enterprise workflow tasks.
Salesforce CEO Marc Benioff Hints at Major Slack AI Coding Features
Marc Benioff also hinted that Salesforce is preparing new AI-powered coding capabilities connected to Slack, the workplace messaging platform Salesforce acquired for $27.7 billion in 2021.
According to Benioff, Salesforce is developing tools designed to lower the barrier to coding for ordinary users. While he did not reveal full details, the Salesforce CEO suggested that new Slack AI coding experiences are coming soon.
“You’re going to see some cool stuff with Slack and code,” Benioff said during the podcast conversation.
The comments suggest Salesforce may be preparing AI agent systems that allow users to create applications, automations, or workflows directly from conversational prompts inside Slack. This could dramatically expand the number of non-technical users capable of building software without traditional programming knowledge.
The rise of no-code and low-code AI systems is becoming one of the biggest themes in enterprise software. Companies including Salesforce, Anthropic, OpenAI, Microsoft, and Google are all competing to dominate AI-powered workplace productivity.
Marc Benioff Wants an AI “Intermediary Layer” for Smarter Token Routing
Another major point from the Salesforce CEO involved AI infrastructure optimization.
Marc Benioff argued that companies should not route every employee request through expensive frontier AI models. Instead, he proposed an “intermediary layer” capable of intelligently routing tasks to different models depending on complexity.
Under this approach, advanced reasoning tasks could go to Anthropic Claude or another frontier model, while simpler requests could be handled by smaller, cheaper AI systems.
The Salesforce CEO suggested this routing system would significantly reduce enterprise AI costs while maintaining strong performance.
This idea reflects a growing industry trend toward multi-model AI architectures. Large companies increasingly want systems that balance performance, cost, speed, and efficiency rather than relying entirely on one expensive frontier model.
As AI token costs continue climbing, enterprise leaders are becoming more focused on optimization strategies that can reduce spending without sacrificing productivity.
Salesforce’s Investment in Anthropic Goes Beyond Token Spending
The relationship between Salesforce and Anthropic extends far beyond AI token purchases.
Salesforce also owns roughly 1% of Anthropic through multiple investment rounds beginning with Anthropic’s Series C funding round in early 2023. Reports indicate Salesforce has invested more than $300 million into Anthropic overall.
This means Salesforce benefits both as a customer and as an investor as Anthropic continues growing inside the enterprise AI market.
Anthropic has rapidly emerged as one of the most important AI companies in the world, competing directly with OpenAI, Google DeepMind, and other major AI developers. Its Claude models have become especially popular among developers and enterprise clients because of their strong coding performance and large context capabilities.
The deeper partnership between Salesforce and Anthropic highlights how enterprise software companies are increasingly aligning themselves with leading AI labs to secure long-term access to frontier AI technology.
AI Agents Are Reshaping Salesforce’s Workforce
Marc Benioff previously revealed that AI productivity gains have already impacted Salesforce’s workforce structure.
Last year, the Salesforce CEO said the company’s support organization headcount dropped from 9,000 employees to around 5,000 employees due to productivity improvements from AI agents.
The comments sparked major debate across the tech industry because they showed how AI automation is beginning to reshape employment inside large corporations.
Support, customer service, coding, sales assistance, and marketing automation are among the first categories being heavily transformed by generative AI systems.
While companies argue that AI improves efficiency and reduces repetitive work, critics continue raising concerns about workforce reductions and long-term employment impacts across the technology sector.
Salesforce, Anthropic, and the Future of AI Coding
The statements from Marc Benioff reinforce how AI coding agents are becoming one of the most valuable and fastest-growing areas in artificial intelligence.
Salesforce spending $300 million on Anthropic tokens demonstrates the enormous scale of enterprise AI adoption now taking place behind the scenes. Large companies are no longer experimenting cautiously with AI. Instead, they are integrating AI agents directly into software engineering, operations, customer support, and internal productivity systems.
The partnership between Salesforce and Anthropic may also signal where the broader AI industry is heading next: AI systems deeply embedded into workplace platforms, collaborative software, and everyday enterprise workflows.
For developers, businesses, and investors, Marc Benioff’s comments are another sign that AI coding is rapidly becoming one of the defining technologies of the modern software era.