Salesforce CEO Marc Benioff is taking a contrarian stance on AI infrastructure spending, arguing that his company’s more conservative approach will ultimately prove superior to the massive capital expenditures being made by tech giants like Meta, Microsoft, and Google.
In a Monday appearance on the “On with Kara Swisher” podcast, Benioff explained that Salesforce isn’t matching the lavish AI spending of its competitors—and he sees this as a strategic advantage rather than a weakness. “I think they’re going to keep spending and it’s going to be expensive for them and it’s going to drive their margins down,” Benioff stated, adding that he plans to “take advantage of their spending to make my products better and lower cost and easier for my customers.”
Salesforce’s AI portfolio includes several enterprise products: Agentforce, a newly announced suite of AI agents that automate workplace tasks; AI Cloud, which hosts large language models from partners like Amazon Web Services; and Einstein Copilot, a generative AI-powered assistant for customer relationship management.
Benioff attributes Salesforce’s cost efficiency to its architectural approach and strategic partnerships. “The way we’ve architected our platform” allows the company to spend less while maintaining highly efficient AI models, he explained. Salesforce relies heavily on third-party data centers from Amazon, Google, and others rather than building extensive proprietary infrastructure. “We tend to use other people’s data centers… and not rely on too much of our own hardware,” Benioff said.
The CEO characterized some competitors’ AI spending as “excessive,” warning it’s “becoming a race to the bottom for some of these companies.” He pointed to energy consumption as evidence of this overspending, noting that several tech giants have invested in nuclear energy deals to power their data centers—“an unusual development,” in his words.
Indeed, Meta recently issued a request for proposals from nuclear energy developers, while Microsoft signed a 20-year agreement with Constellation Energy to restart part of Three Mile Island. Google announced a deal to purchase nuclear energy from small modular reactors built by Kairos Power.
Despite Wall Street’s pressure for AI returns on investment, spending shows no signs of slowing. Meta expects “significant acceleration in infrastructure expense growth” in 2025, while Microsoft’s CFO projected increased capital expenditure driven by cloud and AI demand.
Salesforce’s Q3 research and development costs totaled $1.35 billion, representing a 12.6% year-over-year increase—modest compared to competitors’ infrastructure investments. Perplexity AI CEO Aravind Srinivas echoed Benioff’s sentiment, stating his startup chose to build on others’ models rather than fund proprietary development, which requires “losing billions of dollars a year.”
Key Quotes
I think they’re going to keep spending and it’s going to be expensive for them and it’s going to drive their margins down. I’m going to take advantage of their spending to make my products better and lower cost and easier for my customers.
Marc Benioff, Salesforce CEO, explained his strategy of leveraging competitors’ massive AI infrastructure investments rather than matching their spending. This statement encapsulates his contrarian approach to AI development and his belief that capital efficiency will ultimately prove more valuable than proprietary infrastructure.
We do things a little differently, the way we train, the way we write the software. And also we tend to use other people’s data centers, so we will use Amazon and Google and others and not rely on too much of our own hardware.
Benioff detailed Salesforce’s architectural philosophy, emphasizing partnerships with cloud providers over building proprietary data centers. This approach represents a fundamentally different strategy from competitors investing billions in their own AI infrastructure.
We had a conviction that, number one, models are going to get increasingly commoditized and if you do want to be one of those players that are a provider of the models, you need to have an insane amount of funding and you need to be a company that is losing billions of dollars a year and it’s still fine.
Aravind Srinivas, CEO of Perplexity AI, supported Benioff’s perspective by explaining why his startup chose to build on existing models rather than develop proprietary ones. This reinforces the emerging view that AI model development may not be economically viable for most companies.
While there is a big movement of a lot of companies into these kind of public clouds I think that we have to be careful exactly how much we’re investing.
Benioff cautioned against excessive AI infrastructure spending, suggesting that the current investment frenzy may not be sustainable or strategically sound. This warning comes as Wall Street increasingly questions when massive AI investments will generate returns.
Our Take
Benioff’s position represents a calculated bet on AI commoditization that could reshape industry economics. His argument essentially posits that AI infrastructure will become utility-like, making proprietary data centers a liability rather than an asset. This mirrors historical technology shifts where early infrastructure leaders eventually faced margin compression.
However, there’s risk in this strategy. Companies controlling their own infrastructure gain advantages in customization, data privacy, and long-term cost control. If AI models don’t commoditize as quickly as Benioff predicts, or if proprietary architectures create sustainable competitive moats, Salesforce could find itself dependent on competitors’ platforms.
The nuclear energy investments by Meta, Microsoft, and Google suggest these companies view AI infrastructure as a decades-long strategic asset, not a short-term expense. This fundamental disagreement about AI’s future economics will likely define competitive dynamics for years to come, making Benioff’s contrarian stance one of the industry’s most important strategic debates.
Why This Matters
This story highlights a fundamental strategic divide emerging in the AI industry between companies pursuing vertical integration with massive infrastructure investments versus those leveraging existing platforms for cost efficiency. Benioff’s position challenges the prevailing narrative that AI leadership requires unprecedented capital expenditure, suggesting an alternative path focused on architectural efficiency and strategic partnerships.
The debate has significant implications for AI market dynamics and profitability. If Benioff proves correct, companies like Meta, Microsoft, and Google could face margin compression despite their technological advantages, while more capital-efficient players could deliver competitive AI products at lower costs. This could democratize enterprise AI access and reshape competitive dynamics.
The nuclear energy angle underscores AI’s massive energy demands and environmental impact, raising questions about sustainability and infrastructure requirements. As AI workloads grow exponentially, energy sourcing becomes a critical competitive factor. This story also signals potential market consolidation, where smaller AI companies increasingly rely on infrastructure built by tech giants, creating new dependency relationships and business models in the evolving AI ecosystem.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources: