Key AI and SaaS Trends Dominating Enterprise Tech in 2025

AI Costs and ROI Take Center Stage

Enterprises are moving beyond the hype of AI pilots and demanding tangible returns. A MIT Nanda survey citing a 95% failure rate for AI projects—despite methodological flaws—has become a rallying point for skepticism. Vendors like AWS, ServiceNow, Salesforce, and Workday are pivoting to practical, ROI-driven AI solutions. Workday Executive Chairman Aneel Bhusri noted: "People have gotten a lot more realistic about what AI can do and what AI can't do."

Smaller LLMs Gain Traction

With AI inference costs rising, enterprises are eyeing smaller, more efficient models. Companies like Writer, IBM, and ServiceNow are championing lightweight LLMs such as Palmyra Mini. HubSpot emphasized that not every use case requires the latest LLM, signaling a shift toward cost-effective alternatives.

The Platform Wars Intensify

Every SaaS vendor wants to be a platform, but customer tolerance is limited. Workday mentioned "platform" 78 times during its investor day, while ServiceNow executives said it 21 times in an hour-long earnings call. Salesforce, meanwhile, is expanding beyond CRM into ITSM with its Agentforce platform. The message is clear: platforms are the future, but not everyone will succeed.

Agentic AI as the New UI

Workday’s $1.1 billion acquisition of Sana underscores the bet on AI as the primary user interface. Gerrit Kazmaier, Workday’s president of product and technology, compared Sana to "the iOS for enterprise," envisioning it as the "front door for work." The goal is to replace traditional SaaS layouts with AI-driven interactions, which users currently engage with 7 times a day.

Build vs. Buy Debate Rages On

Enterprises face a dilemma: build custom AI solutions or buy off-the-shelf? Workday and Salesforce argue that DIY AI is fraught with challenges, from unready data to complex processes. As Kazmaier put it, "You cannot just superficially slap AI across legacy APIs and hope it transforms the process." The consensus? Midmarket firms will buy, large enterprises will do both, and startups will build.

Is an AI Bubble Inevitable?

Signs of an AI bubble are mounting, from soaring valuations to the return of SPACs. OpenAI executives have openly acknowledged the risks, drawing parallels to the dot-com crash. As one Constellation Research article asked: Are we repeating history?

Tech Debt and Code Transformation

Oracle and MongoDB are betting on AI-driven code transformation to modernize legacy systems. Oracle CTO Larry Ellison highlighted the dual advantage of building both application generators and applications. However, tech debt is a relentless cycle—modernization today breeds new debt tomorrow.