Leaders from Dell, Microsoft, Salesforce, ServiceNow, and Snowflake have released their 2026 predictions for AI in the workplace, and they agree that safeguards for AI agents and ROI are the top priorities for their customers.

Dell CTO John Roese sees the future of AI living on premises where organizations have more control over security, governance, and cost. He said AI tools like agents and chatbots have been rushed into production without sufficient policies in place.

“This is not just risky; it’s unsustainable,” he writes. “By 2026, the demand for robust frameworks and private environments to ensure stability and control will be undeniable. Running models locally—on-premises or in controlled AI factories—will become the norm to provide a stable foundation and insulate organizations from external disruptions. But this is more than a prediction. It’s an urgent appeal.”

While Roese said solid ROI has begun to emerge among large companies capable of integrating AI, ServiceNow said in 2026 AI will be defined by the value it brings to the bottom line.

“That’s the only question that matters,” said Heath Ramsey, ServiceNow’s group vice president of AI product platform management in his video predictions. “Leaders face a simple directive to turn AI investments into measurable value, fast.”

Ramsey urged companies to start small and look for the task that is “bleeding time and money,” then fix it end-to-end with AI. But to make that effective, he turned the conversation back to governance.

“You need one entry point with clear policies and approvals,” he said. “A secure foundation for your AI is how you turn an isolated win into a repeatable pattern that scales across your business.”

In terms of governing AI agents, Snowflake CISO Brad Jones says the job in front of his counterparts is to walk the line between putting up guardrails around their behavior while leaving room for operators to experiment and innovate, which is a job for data governance.

“There are likely to be many documents or data sets in a company that don’t have permissions correctly locked down,” he says, in the company’s 2026 AI predictions. “If you feed that into generative or agentic AI, the tool may expose data that it shouldn’t.”

It's not just internal productivity at stake either, according to Saleforce. It is each company's image and reputation in the marketplace, predicts Adam Evans, EVP & GM, Salesforce AI.

“By 2026, brands won’t be defined by logos or slogans; they will be defined by their AI. These customizable agents will become the ultimate brand ambassadors: smart, personalized, and continuously evolving with every exchange. In this new reality, the divide will be absolute, and the brands that win will be the ones whose AI delivers a consistently exceptional experience, and everyone else will fall behind.” .

Microsoft sees trust as the essential ingredient to deploying AI agents correctly, according to its 2026 prediction.

Vasu Jakkal, corporate vice president of Microsoft Security, said for humans to rely on agents to carry out tasks or for decision-making, the agents need a clear identity, limits on accessing systems, protocols for managing data they create, and ways to protect that information from attackers.

“Every agent should have similar security protections as humans, to ensure agents don’t turn into ‘double agents’ carrying unchecked risk,” she said in the prediction paper.

The largest software company in the world also offered a hot take on hardware, saying AI growth is not just dependent on who has the biggest data center. Industry veteran Mark Russinovich, chief technology officer for Microsoft Azure, said 2026 is about putting compute to work efficiently.

“The most effective AI infrastructure will pack computing power more densely across distributed networks,” Russinovich says.

• Microsoft teases agents that become 'independent users within the workforce'
• Linux Foundation aims to become the Switzerland of AI agents
• How to answer the door when the AI agents come knocking
• AI agents get office tasks wrong around 70% of the time, and a lot of them aren't AI at all

He predicts that 2026 will see “the rise of flexible, global AI systems — a new generation of linked AI 'superfactories' — that will drive down costs and improve efficiency.”

Microsoft unveiled the first of its linked AI superclusters in Wisconsin this September, which boasts 1.2 million square feet of floor space and thousands of interconnected Nvidia GB200s in a 337 MW barn capable of processing 865,000 tokens per second. It plans to join that facility with other AI superfactories it builds as those come on line.

Back at Dell – the global market leader in the sale of x86 servers and numerous categories of high end commercial data storage used in AI systems – Roese said the widespread deployment of AI hardware redraws the goal lines for resilience and disaster recovery in 2026.

“The focus shifts from simply backing up systems to ensuring AI capabilities remain functional, even if primary systems go offline,” he predicts. “This involves protecting vectorized data and other unique AI artifacts, allowing the intelligence of the system to persist through any disruption. Achieving this requires innovation across the entire AI value chain – from data protection and cybersecurity companies to core AI technology providers.” ®