As CxOs zoom out on 2025, it's clear that the year was characterized by AI building blocks and the need for real returns. Agentic AI platforms aren't quite mature, but the industry standards and connections are making deployments more realistic.

If you're an enterprise technology buyer the onus is on the vendors for showing value and more than proof of concepts. The game for both the buy side and the sell side revolves around use cases that show returns and can scale across the enterprise.

For vendors, the appeal of this use-case-by-use-case motion is selling a platform that can create, manage and orchestrate AI agents. For CxOs, the chase is the autonomous enterprise and efficiency and productivity gains that can fund more AI efforts.

In many respects, 2025 was a tale of two halves. The first half was dominated by economic volatility and skepticism about what vendors were pitching.

In the second half, the vendor platforms matured and proofs of concepts began to move to production. There also seems to be consensus that AI, automation and process are comingled.

Get Constellation Insights

For the 2025 analysis, I homed in on all the Constellation Insights articles posted throughout the year and analyzed the broad themes in enterprise technology, the buy side and the sell-side vendors.

Common themes in 2025

Agentic AI grew up, but still isn't what any reasonable executive would call mature. Agentic AI is a heavy lift that includes forward-deployed engineers, an enterprise data strategy that works, services and use case refinement. Simply put, it's easy to build an AI agent. Scaling them with guardrails and building enough trust to let these models run your company is another matter entirely.

AI economics in flux. Salesforce's recent move to offer an agentic enterprise license agreement is notable because it gives enterprises predictability. The consumption approach from SaaS didn't work out for many buyers. Vendors want to monetize the value they're creating, but enterprises are tired of bills continuing to rise. A consensus emerged that there will be multiple models that are combined to optimize for price performance. Hyperscalers are working custom silicon to commoditize compute and models will be rightsized for the task at hand.

Everything is a platform. SaaS providers are branching out beyond their core markets to become broader AI agent platforms. There isn't a vendor that doesn't have a horse in the AI agent platform race.

The real wins will revolve around industry-specific use cases and first-party data. Enterprises are beginning to leverage their unique data to train models and optimize use case by use case and industry by industry. Context is everything. Again, it's easier said than done.

The buy side

AI is an excuse to change operating models. Enterprises are beginning to use the agentic AI push as a way to optimize operations across all processes and use cases. These efforts would take longer under yesterday's transformation projects, but AI is giving smart companies air cover to accelerate plans.

Models are a commodity. Enterprises are no longer wowed by the latest and greatest models. China's DeepSeek and Qwen from Alibaba changed that proprietary model thinking in early 2025. Now enterprises are just as likely to use Nova models from AWS or open source as they are the big three models. That said models are a key ingredient of buying decisions of broader platforms.

Multi-cloud, multi-model, multi-everything. Enterprises have always been wary of lock-in (and still do it anyway), but AI is advancing so quickly that CxOs want neutrality and options. Interoperability is a requirement and AWS, Google Cloud and Microsoft Azure are beginning to connect. AI sprawl isn't a concern yet, but will be soon.

Technology strategy and business strategy merge. Spend on infrastructure, applications, models and ecosystems are driven by margins, agility and resilience of all types. Enterprises will look to AI as a way to abstract away technical debt.

Value over vision. Enterprises are shrinking budget cycles and demanding measurable impact. Enterprises care about efficiency, productivity, customer experiences and revenue gains. Vanity pilots are kaput. Practical AI is in.

It's still all about data. The enterprises that have a coherent data strategy, data quality and modern pipelines and platforms can win in the AI age. Too few enterprises have their data game down.

The sell side

Every vendor wants to be your sole platform. A tenured CxO would chuckle at these AI vendor developments. Why? Vendors have always wanted to consolidate all of your spending with them. Picking a vendor that can be your one go-to agentic AI platform isn't easy.

The value sales pitch. Vendors increasingly talked enterprise value, use cases and process optimization. They're certainly talking a good game. By mid-2025, this value-based sales pitch became the norm. Enterprise vendors and even OpenAI and Anthropic are all using the industry and use case playbook with partners or via direct sales.

LLM giants build ecosystems. OpenAI and Anthropic, not to mention Google's Gemini, are building ecosystems and applications that can leverage their models. What's unclear is whether these LLM giants can be the overarching interface that relegates enterprise software to plumbing. Also keep an eye on Databricks and Snowflake, which could leverage their data platforms to be the overarching enterprise layers.

Orchestration is the big thing. Vendors are pitching themselves as your go-to AI orchestration layer. For a vendor, agent orchestration will drive stickiness and revenue and lock-in. It's also possible that orchestration and data gravity will go together and drive vendor revenue.

Hyperscalers gain clout. Aside from ServiceNow, hyperscalers and integrators are the most natural fit to be that overarching AI platform, abstraction and orchestration layer. The big three cloud players are horizontal, touch every part of the enterprise and have pricing models that naturally go together with AI agents. Add it up and AWS, Microsoft Azure and Google Cloud all gained clout in 2025.

The year in Insights

Business Research Themes

Data to Decisions, Future of Work, Innovation and Product-led Growth, New C-Suite, Next-Generation Customer Experience

Roles

Chief Executive Officer, Chief Financial Officer, Chief Information Officer

Tags

Artificial Intelligence (AI), AI business transformation