Latest
Tuesday, November 17, 2026
  • New YorkNY
  • LondonLDN
  • TokyoTYO
Business

Salesforce Implements AI Spending Limits to Control Costs

The company's new policy allocates monthly token budgets for every role, allowing employees to focus their AI spend on high-priority tasks.

Salesforce sets AI budgets after employees spent 'thousands of dollars per month on tokens,' memo says
Source: Business Insider

Salesforce is taking steps to rein in its AI spending after some employees spent thousands of dollars per month on tokens last quarter.

According to an internal memo viewed by this publication, the company's Chief Operating Officer Miguel Milano has implemented a new policy that allocates monthly token budgets for every role, allowing employees to focus their AI spend on high-priority tasks.

The move reflects a common challenge faced by companies embracing AI technology - balancing cost control with employee empowerment. Other businesses have introduced similar spending limits in an effort to curb the rising costs of artificial intelligence.

While Salesforce is setting AI budgets at the team level, rather than companywide, team leaders will be responsible for managing their groups' AI usage. The goal, as stated by a company spokesperson, is to ensure that AI investments deliver maximum impact for both business and customers.

Salesforce has tested this new approach on one team with promising results, reporting that the group operated well under budget after just two months without any decline in productivity.

The sales organization under President and Chief Revenue Officer Alexa Vignone will have a strict $1,000 monthly limit for technical roles, while non-technical employees like sellers are allocated just $100 per month for AI tokens.

In contrast, the Customer Success team led by Jim Roth has a more modest budget of $500 per month for technical personnel, with non-technical staff receiving the same $100 allowance as their counterparts in other departments. Meanwhile, the Global Professional Services and Forward Deployed Engineering teams under Lori Steele are also subject to these new spending limits.

Other companies have been experimenting with similar measures to control AI expenses, including Coinbase which has implemented weekly caps ranging from $500 to $5,000 based on employee roles and seniority. JPMorgan Chase, for instance, has introduced a monthly limit of $2,000 for some engineers using Anthropic's Claude Code.

The company's CEO Marc Benioff had previously estimated that Salesforce would spend around $300 million on Anthropic tokens this year, primarily to develop coding agents. This new approach aims to focus AI spending on projects with the greatest impact for customers and the business, rather than simply encouraging adoption.

To manage AI spending, Salesforce has introduced new budget limits for employees based on their roles within the company.

Monthly token spend limits have been set at varying levels depending on an employee's role, with technical roles such as Alexa's Org Technical Roles allowed to spend up to $1,000 per month, while non-technical GTM/Seller Roles are capped at $100 per month.

This new system aims to reduce the financial burden of AI spending, which had previously seen employees exceed their budgets by thousands of dollars each month. The company is now encouraging employees and their managers to work together to stay within allocated limits.

As part of this initiative, leaders will be responsible for managing their team's budgets, much like they would with travel and entertainment expenses.

The new AI budget guidelines specify that any request exceeding 3 times the individual limit will require approval from a more senior manager, namely Alexa, Jim, or Lori, who will also involve finance in the decision-making process.

These higher-value requests are reviewed on a monthly basis to ensure they align with company goals and justify the additional expenditure through a clear return on investment justification.

The increased budget will be valid for 90 days before being reassessed. This temporary allocation is intended to support ongoing projects, but it's not a long-term solution. A more comprehensive process is in development to better manage AI spend.

As part of this interim setup, leaders are encouraged to track individual and team usage through the AI Spend Dashboard. They should review usage monthly to identify areas where costs can be optimized without compromising results.

The memo advises employees to regularly review their AI budget usage to identify areas where costs can be optimized without sacrificing results.

Visibility into pooled team spending will be provided for leaders in countries represented by the Works Council, allowing them to monitor and manage token allocation effectively.

With visibility comes responsibility; while individual employee usage will not be directly visible to leaders, they are expected to work with their teams on developing best practices for token management.

Facts based on reporting originally published by Business Insider.

You may republish this story, in full or in part, if you credit News Central Site and link to it (licence CC BY 4.0). Photos are not included.