Gartner: AI models and platforms spending to hit $64B in 2026 as companies shift focus to governance

Global end-user spending on AI models and AI platforms is expected to reach $64 billion in 2026—up 63.4% from 2025—according to new forecasts from Gartner. The research firm says the sharpest jump will come from generative AI (GenAI) model spending, while AI platforms grow at a slower but still brisk pace.

Behind the headline growth, Gartner frames the central enterprise question less as whether to use AI and more as where—and how—to deploy it without losing control of costs, data, and day-to-day usage.

$64 billion in 2026, up 63.4% from 2025, Gartner says

Gartner projects worldwide end-user spending on AI models and AI platforms will total $64 billion in 2026, up from $39 billion in 2025. Gartner positions the forecast as a read on real enterprise demand, breaking the market into spending on models and spending on platforms used to train, deploy, monitor, and administer AI inside corporate IT environments.

The expected surge comes as most large organizations have already run pilots and are now being pushed by top leadership to move toward industrial-scale rollouts with more structured budgets. Gartner’s view is that decisions are increasingly driven less by the technology’s promise than by the ability to scale, integrate AI into existing processes, secure data flows, and avoid operational blowups—financial or regulatory.

Gartner also stresses that this is not a uniform market. It draws a line between fast-accelerating GenAI model spending and platform spending tied to infrastructure, governance, and orchestration. The widening gap between those segments, Gartner argues, signals rising demand both for models tuned to specific use cases and for the tools that keep enterprise AI use under control.

Gartner notes the $64 billion figure is narrower than other Gartner estimates that cite much higher AI spending when hardware, services, and related software are included. This forecast focuses specifically on models and platforms—smaller in scope, but strategically important because they determine whether AI projects can be industrialized and whether vendors can secure durable positions inside organizations.

GenAI model spending seen up 117%; AI platforms up 36.9%

Gartner forecasts GenAI model spending will grow 117% in 2026. The firm describes that jump as a sign companies are moving beyond experimenting with generic assistants and instead buying access to models, inference capacity, customization options, and vendor-provided security and traceability features.

At the same time, spending on AI platforms is expected to rise 36.9%. Gartner’s platform category includes runtime environments, data pipelines, MLOps and LLMOps tools, model version management, observability, oversight, and in some cases access-control mechanisms. The growth rate is slower than models, but Gartner still calls it high—reflecting the need for stable foundations as deployments move into production across on-premises, cloud, and hybrid environments.

On the ground, Gartner says the split shows up in how companies buy. Many organizations fund model access first to unlock quick wins—customer support, document search, internal content generation, or software development assistance. But those deployments can quickly create side effects: a rapid spread of use cases, sensitive data moving around, vendor dependence, variable inference costs, and compliance demands.

That’s why platforms are increasingly both a defensive and offensive line item, Gartner argues. Defensive, because they help limit risk and standardize practices. Offensive, because they speed production rollouts and make it easier to reuse assets—prompts, connectors, specialized models, datasets—across business units. Gartner describes this as a race toward “platformization,” with major cloud providers, software publishers, and specialists all trying to make their tools the central layer for enterprise AI.

Specialized “DSLM” models forecast to jump 210%, reshaping procurement

Among the fastest-growing categories, Gartner highlights DSLMs—domain-specific language models—and specialized models more broadly, which it expects to grow 210% in 2026. The pitch is a move away from general-purpose approaches toward more reliable performance in specific domains such as law, health care, finance, industry, customer relations, and cybersecurity.

Gartner ties that shift to limits companies have encountered with general-purpose models. Enterprises want higher accuracy with their own vocabulary, fewer hallucinations, more predictable behavior, and clearer assurances about the provenance and scope of training data. In some sectors, Gartner adds, the push for controlled models also intersects with sovereignty and data-localization constraints, as well as internal validation requirements before production deployment.

In procurement, Gartner says the trend could change the nature of RFPs. Where buying a general-purpose model can resemble subscribing to a service, acquiring a specialized model often requires adaptation work, data governance, business testing, and ongoing supervision. Contracts may also include performance commitments, security measures, retention policies, and portability clauses.

Gartner also points to a growing number of players. Established cloud and AI vendors are offering their own specialized models or partner catalogs. Vertical software publishers are trying to monetize models trained on industry-specific corpora. And some companies are considering building internal models when their data is a competitive advantage. The economics vary, Gartner notes, because design and maintenance can be expensive—but the promise is lower usage costs at scale and tighter control over outputs.

Gartner’s key message: governance will decide the winners

Beyond the spending totals, Gartner emphasizes governance—who can use AI, where it runs, and how it’s applied across the business. In Gartner’s communication, analyst Cheparthi said “the biggest winners” will be vendors that can help enterprises manage where and how AI is used, a framing that points to broad AI governance: use-case inventories, access controls, data policies, compliance, and financial oversight.

That governance push is highly operational. Gartner says organizations are trying to avoid a sprawl of unmanaged tools—sometimes adopted directly by business units—and to centralize rules around permitted data types, confidentiality levels, activity logs, and model evaluation mechanisms. The goal is twofold: reduce legal and reputational risk, and prevent cost overruns tied to usage-based model billing.

Gartner also links governance demand to the rise of monitoring and assurance tools: automated testing, response-quality evaluation, data-leak detection, red teaming, and version management. Those capabilities are increasingly built into platforms or sold as dedicated solutions. For enterprises, Gartner says the challenge is choosing an architecture that supports fast innovation and control—without blocking business teams or piling on redundant technical layers.

In that environment, Gartner suggests the market may consolidate around a few selection criteria: the ability to orchestrate multiple models, compatibility with hybrid environments, transparency into performance and cost metrics, and integration with existing security frameworks. Vendors that can demonstrate concrete value—repeatable deployments, measurable incident reduction, and more predictable budgets—stand to capture a significant share of the growth Gartner expects in 2026.

Frequently asked questions

What exactly does Gartner’s $64 billion 2026 forecast measure? It targets end-user spending dedicated to AI platforms and AI models—the tools and services needed to access models, use them, and administer them—rather than total AI spending including hardware, services, and adjacent software.

Why would GenAI model spending rise faster than platform spending? Gartner says GenAI models are often purchased to quickly unlock use cases through usage-based services or subscriptions. Platforms then ramp up to industrialize, secure, and standardize those uses, but platform projects typically take longer to integrate and govern.

What are DSLMs? DSLMs are domain-specific language models designed to deliver better performance and reliability in a given sector or business vocabulary, with the goal of reducing errors and improving fit with compliance constraints.

What criteria could separate vendors, according to Gartner? Gartner emphasizes the ability to help enterprises manage where and how AI is used—covering governance, multi-model orchestration, data controls, monitoring, compliance, and cost management tied to model usage.

Key takeaways

Gartner projects $64 billion in 2026 spending on AI platforms and AI models, up 63.4%.

GenAI model spending is forecast to rise 117% in 2026, versus 36.9% for AI platforms.

Specialized models and DSLMs are expected to be the fastest-growing category, up 210%.

AI governance—controlling where and how AI is used—is becoming a central market differentiator.

Sources

Gartner press release: “Gartner Forecasts Worldwide AI Platforms and Models Market to Grow 63% in 2026.”

Gartner post on X: “Gartner Forecasts Worldwide AI Platforms and Models …”

ELE Times: “Gartner Forecasts Worldwide AI Platforms and Models Market to Grow 63% in 2026 …”

ChannelNews.fr: “Gartner voit les dépenses mondiales d’IA bondir à 2.600 …”

Distributique: “2 500 Md$ de dépenses IA en 2026 … selon Gartner.”