Enterprise AI has reached an inflection point. In just two years, enterprise spending on generative AI surged from $1.7 billion to $37 billion—a 22x increase that now represents 6% of the global SaaS market. According to Menlo Ventures' 2025 State of Generative AI in the Enterprise report, this makes generative AI "the fastest-scaling software category in history."
But the raw spending numbers only tell part of the story. Beneath the surface, the data reveals a fundamental restructuring of how enterprises build, buy, and deploy AI—and a widening gap between organizations capturing value and those still searching for it.
Drawing on comprehensive research from OpenAI's State of Enterprise AI report, Menlo Ventures' survey of 495 enterprise decision-makers, and supplementary data from a16z and other sources, this analysis examines where the enterprise AI market stands at the end of 2025—and what it reveals about the future of business.
The Scale of the Shift
The numbers demand attention. Enterprise AI didn't just grow—it exploded.
According to Menlo Ventures' analysis:
- $37 billion in total enterprise AI spend in 2025, up from $11.5 billion in 2024 (3.2x year-over-year growth)
- The median sector grew 6x or more in AI spending
- Technology sector AI spending increased 11x year-over-year
- Enterprise AI now represents 6% of the global SaaS market
OpenAI's data corroborates this acceleration from the consumption side:
- ChatGPT message volume grew 8x year-over-year
- API reasoning token consumption increased 320x per organization
- 7 million+ ChatGPT workplace seats deployed
- Enterprise seats grew 9x year-over-year
This isn't incremental adoption. It's a phase change in how enterprises operate.
The speed of this shift is historically unprecedented. As Menlo Ventures notes, enterprise AI's growth trajectory "eclipses any software category we've tracked."
Where the Money Is Going
Understanding where enterprises are investing reveals the priorities shaping the market.
The Infrastructure-Application Split
Enterprise AI spending divides roughly in half:
- $19 billion (51%) on AI applications
- $18 billion (49%) on infrastructure
Within infrastructure, the allocation reveals where enterprises see foundational value:
- $12.5 billion on foundation model APIs
- $4.0 billion on model training infrastructure
- $1.5 billion on AI infrastructure and data management
The Application Layer
The $19 billion application market breaks into three categories:
Departmental AI ($7.3 billion): Function-specific tools, with coding dominating at $4 billion (55% of departmental spend)
Horizontal AI ($8.4 billion): Cross-functional tools like enterprise copilots, which hold 86% share of horizontal AI spend
Vertical AI ($3.5 billion): Industry-specific solutions, with healthcare leading at $1.5 billion
The concentration is notable: coding tools and enterprise copilots together account for nearly two-thirds of application spending.
Coding: AI's First "Killer Use Case"
If any category demonstrates AI's enterprise potential, it's coding. The numbers are staggering.
According to Menlo Ventures:
- The AI coding market reached $4 billion in 2025
- This is up from $550 million in 2024—a 7.3x increase
- 50% of developers now use AI coding tools daily
- In top-quartile organizations, that number rises to 65%
- Teams report 15%+ velocity gains on average
OpenAI's data adds texture: organizations using their tools for coding saw reasoning token consumption increase 320x per organization as teams moved from simple completions to complex, multi-step programming tasks.
Perhaps most striking: one CTO quoted in the Menlo Ventures report noted that "90% of our code is now AI-generated"—up from 10-15% a year ago.
Coding has become AI's first true "killer use case" in the enterprise—a category where adoption is near-universal among leading organizations and productivity gains are measurable and significant.
The Model Wars: Anthropic's Rise
One of the most dramatic shifts in 2025 occurred in the competitive landscape for large language models. The data reveals a significant reordering of market share.
Enterprise LLM Market Share
| Provider | 2023 | 2024 | 2025 | Change (2023-2025) |
|----------|------|------|------|---------------------|
| Anthropic | 12% | 24% | 40% | +28 percentage points |
| OpenAI | 50% | 34% | 27% | -23 percentage points |
| Google | 7% | 16% | 21% | +14 percentage points |
Anthropic's ascent is particularly notable. The company more than tripled its enterprise market share in two years, moving from distant third to clear leader.
Coding Market Share
The shift is even more pronounced in coding, where Anthropic now dominates:
- Anthropic: 54% of coding market share
- OpenAI: 21%
- Others: 25%
What explains Anthropic's rise? Multiple factors converged: the launch of Claude Sonnet 3.5 in June 2024 delivered a capability jump that developers quickly noticed, and the release of Claude Code further solidified Anthropic's position in the coding workflow. Enterprise buyers followed developer preferences.
This market restructuring carries strategic implications. When the dominant platform shifts this rapidly, enterprises must consider not just current capabilities but trajectory—and what it means for their AI investments.
Buy vs. Build: The Decisive Shift
Perhaps no trend better captures enterprise AI's maturation than the shift from building to buying.
In 2024, enterprises split their AI development almost evenly:
- 47% built internally
- 53% purchased
By 2025, the balance shifted dramatically:
- 24% built internally
- 76% purchased
This 52 percentage point swing toward purchasing reflects several dynamics: maturing vendor solutions, pressure to deploy faster, and growing recognition that most AI capabilities don't require custom development.
The PLG Revolution
Product-led growth (PLG) is transforming how enterprises acquire AI:
- 27% of AI application spend now flows through product-led growth channels
- This is 4x the rate of traditional software (7%)
- Including shadow AI, PLG may represent ~40% of total AI spend
- AI deals convert at 47% versus 25% for traditional SaaS—nearly 2x the rate
The implications are significant. When employees can adopt AI tools without procurement approval, buying patterns shift. Enterprise IT teams increasingly find themselves managing AI portfolios that emerged bottom-up rather than top-down.
The Productivity Payoff
Are enterprises seeing returns on their AI investments? OpenAI's survey of 9,000+ workers across functions provides evidence.
Broad-Based Gains
- 75% of workers report improved speed or quality from AI
- Workers save 40-60 minutes per active day on average
- Data science and engineering roles save 60-80 minutes per day
- 75% report completing tasks they previously couldn't
Function-Specific Impact
The gains vary by function, with some areas seeing outsized benefits:
- 87% of IT workers report faster issue resolution
- 85% of marketing professionals achieve faster campaign execution
- 75% of HR teams see improved employee engagement
- 73% of engineers report faster code delivery
These are not marginal improvements. Organizations where AI adoption is widespread are fundamentally changing the productivity equation.
The Growing Divide
Perhaps the most consequential finding across these reports: the gap between leaders and laggards is widening, not narrowing.
The Frontier Gap
OpenAI's data reveals stark differences in how organizations use AI:
- Frontier workers send 6x more messages than median workers
- Frontier firms send 2x more messages per seat and 7x more to custom GPTs
- In coding specifically, frontier workers send 17x more messages than average
- Workers using 7+ task types save 5x more time than those using fewer
The Underutilization Problem
Despite high adoption, significant capabilities remain unused:
- 19% of monthly active users have never used data analysis features
- 14% have never used reasoning capabilities
- 12% have never used search functions
This pattern—broad adoption but shallow utilization—explains much of the value gap. Organizations that achieve deep, varied use of AI capabilities pull away from those that deploy AI narrowly.
The Investment Gap
The divide extends to investment levels:
- More than one-third of high performers allocate over 20% of digital budgets to AI
- Only 7% of other organizations reach this threshold
As McKinsey's 2025 State of AI research confirms, 88% of organizations now use AI, but only 6% capture meaningful enterprise value. The data suggests this gap is structural, not temporary.
What Leaders Do Differently
Across these reports, consistent patterns emerge among organizations capturing the most value from AI:
1. Deep System Integration
Leading organizations don't just deploy AI tools—they connect them. They use connectors and integrations that provide AI with business context, enabling more accurate, relevant responses. Isolated AI tools deliver isolated gains; integrated AI transforms workflows.
2. Workflow Standardization and Reuse
High performers build reusable AI assets: custom GPTs, API assistants, standardized prompts. This "encode once, use many" approach multiplies returns on AI investment. OpenAI's data shows frontier firms use custom GPTs at 7x the rate of average organizations.
3. Executive Leadership and Sponsorship
Organizations with visible executive commitment to AI consistently outperform those where AI remains an IT initiative. According to McKinsey, high performers are three times more likely to report active senior leadership engagement.
4. Data Readiness and Continuous Evaluation
AI performs only as well as the data it accesses. Leading organizations invest in data quality, accessibility, and governance—and continuously evaluate AI performance against business metrics.
5. Deliberate Change Management
Technology deployment without organizational adaptation fails. High performers appoint AI champions, develop training programs, and explicitly manage the cultural and process changes AI requires.
What's Next: 2026 Predictions
Menlo Ventures offers five predictions for the year ahead:
1. AI will exceed human performance in daily practical programming tasks. The coding productivity gains already evident will compound as models improve and developers adapt their workflows.
2. Jevons' paradox continues—net spend rises despite falling inference costs. As AI becomes cheaper per unit, organizations will use dramatically more of it. Total spending will increase even as unit economics improve.
3. Explainability and governance go mainstream for agents. As AI agents take on more autonomous tasks, enterprises will demand transparency into their reasoning and robust governance frameworks.
4. Models move to the edge. On-device and mobile GPU deployment will expand, enabling AI capabilities without cloud dependencies—critical for latency-sensitive and offline use cases.
5. One major use case outside coding achieves widespread adoption. The success pattern established in coding will replicate in another domain—potentially customer service, sales, or content creation.
The Bottom Line
The 2025 enterprise AI data tells a clear story: we've moved from the experimentation era to the infrastructure era. AI is no longer a pilot program or innovation initiative—it's becoming core enterprise infrastructure, with $37 billion in spending to prove it.
But the data also reveals a growing divide. Organizations that invest deeply, integrate thoroughly, and deploy broadly are pulling away from those still experimenting at the margins. The 6x gap in usage between frontier and median workers isn't just a statistic—it's a competitive moat that widens with each passing quarter.
For enterprise leaders, the strategic implications are clear:
- The buy vs. build question is largely settled. With 76% of AI now purchased, custom development should be reserved for genuine differentiation.
- Coding represents a solved category. Organizations not yet using AI coding tools at scale are leaving measurable productivity gains on the table.
- Model choice matters—and it's shifting. Anthropic's rise from 12% to 40% market share in two years demonstrates how quickly the landscape can change.
- Depth beats breadth. Organizations using AI deeply across multiple task types capture 5x more value than those with shallow adoption.
- The investment threshold is rising. With leaders allocating 20%+ of digital budgets to AI, matching their commitment requires matching their resources.
Enterprise AI has moved from experiment to infrastructure. The organizations that treat it accordingly—with serious investment, deep integration, and sustained commitment—will define the next era of competitive advantage.
The only question remaining is which side of the divide your organization will land on.
At OuterEdge, we help organizations close the gap between AI adoption and AI transformation. Our technology products—Growth|Edge and Customer|Edge—combined with our expert services in Sales, Product, and Process Transformation, enable enterprises to capture the full value of AI investment. If you're ready to move from experimentation to infrastructure, book a strategy call to discuss your AI transformation.