The State of Enterprise AI: $37B, 320x, and the Widening Gap

⏱️ Your Weekly Brief

Welcome Back,

This week, we're diving into three reports that paint a clear picture of where enterprise AI stands heading into 2026.

• OpenAI dropped their first comprehensive enterprise report showing 8x year-over-year message growth and a widening gap between AI leaders and laggards.

• Menlo Ventures quantified the market at $37 billion (up 3.2x from last year) and revealed that Anthropic has quietly unseated OpenAI as the enterprise LLM leader.

• McKinsey is making the case that your biggest AI bottleneck isn't technology, it's whether your leaders have built their "second muscle."

Let’s dive in 🎄

🔗 Open AI’s State of Enterprise AI 2025

The GPTLDR

OpenAI's first enterprise-focused report, drawing from over 1 million business customers and survey data from 9,000 workers across nearly 100 enterprises, reveals that enterprise AI has shifted from experimentation to core infrastructure.

The headline? A 320x increase in API reasoning token consumption per organization year-over-year, with workers saving 40-60 minutes per day. But the real story is the widening gap between frontier firms and everyone else.

The Details

Usage is exploding—and deepening

• ChatGPT Enterprise seats increased 9x YoY; weekly Enterprise messages grew 8x

• 20% of all Enterprise messages now flow through Custom GPTs or Projects

Workers are reporting real productivity gains

• 75% of surveyed workers report improved speed or quality

• Time saved per active day: 40-60 minutes (data science and engineering: 60-80 minutes)

• 87% of IT workers report faster issue resolution

• 73% of engineers report faster code delivery

• 75% of users can now complete tasks they previously couldn't perform

Non-technical roles are picking up technical work

• Coding-related messages outside engineering/IT/research grew 36% in six months

• Workers engaging across 7+ task types report 5x more time savings than those using only 4

The leader/laggard gap is real—and widening

• Frontier workers (95th percentile) send 6x more messages than median

• Frontier workers use data analysis tools 16x more than median workers in the same function

• Frontier firms generate 7x more messages to Custom GPTs than median enterprises

• For coding specifically, frontier workers send 17x more messages than median

What Matters

Here’s what leaders are doing differently at Frontier firms.

• Enable connectors for secure data access (25% of enterprises still haven't)

• Standardize workflows through reusable Custom GPTs

• Maintain executive sponsorship and clear mandates

• Invest in data readiness and continuous evaluation

• Deploy deliberate change management with embedded AI champions

🔗 Menlo Ventures State of Generative AI in the Enterprise

The GPTLDR

Menlo's third annual enterprise AI survey puts hard numbers on the market: $37 billion in 2025 spend (up 3.2x YoY), with the application layer capturing $19 billion—over 6% of the entire software market just three years after ChatGPT's launch. The surprise finding: Anthropic now commands 40% of enterprise LLM spend, unseating OpenAI. Coding has emerged as the definitive "killer use case" at $4 billion, and startups are winning the application layer 2:1 over incumbents.

The Details

The build vs. buy equation has flipped

• 76% of AI solutions are now purchased vs. built (was 53% in 2024)

• AI deals convert at 47% vs. 25% for traditional SaaS

• Product-led growth drives 27% of AI app spend (4x the rate of traditional software)

LLM market share shift: Anthropic takes the lead

• Anthropic: 40% enterprise share (up from 24% last year, 12% in 2023)

• OpenAI: 27% (down from 50% in 2023)

• Google: 21% (up from 7% in 2023)

• In coding specifically: Anthropic holds 54% vs. OpenAI's 21%

Coding is the breakout category

• Coding AI spend: $4B (55% of all departmental AI spend)

• 50% of developers now use AI coding tools daily (65% in top-quartile organizations)

• Teams report 15%+ velocity gains

What's not happening yet

• Only 16% of enterprise and 27% of startup deployments qualify as true agents

• Most production architectures remain "basic if-then logic around a model call"

• Prompt engineering still dominates customization; fine-tuning and RL remain niche

What Matters

• The Anthropic/OpenAI reversal should inform your model strategy. If you're locked into a single provider, you're exposed. Multi-model approaches are becoming table stakes.

• The 47% conversion rate for AI deals suggests buyer intent is exceptionally high—but also that procurement processes are being compressed. Traditional sales cycles may be permanently shortened.

• Coding capturing 55% of departmental AI spend signals where proven ROI exists today. If your engineering teams aren't using AI-assisted development, they're at a competitive disadvantage.

🔗 McKinsey: Building the AI Muscle of Your Business Leaders

The GPTLDR

McKinsey argues that the critical bottleneck for AI transformation isn't technology—it's domain leaders. These N-2 and N-3 executives (two to three levels below the CEO) who own end-to-end business processes need a "second muscle": enough tech fluency to develop AI-enabled transformation roadmaps, oversee tech delivery, and lead change management.

The Details

The domain leader gap is severe

• Analysis of 903,565 senior leaders across Fortune 500 companies

• Only 17% of skill sets are technical in nature

• Just 5% of careers included holding a technical role

• Most companies derive the majority of AI benefits from a few deeply transformed domains

• Each domain typically requires 5-15 interrelated use cases to capture real transformational value

What defines an effective domain owner

• Reimagine their domain with customer-centric, AI-enabled vision—not just automate existing workflows

• Develop AI-enabled transformation roadmaps with sequenced use cases and clear KPIs tied to outcomes

• Oversee tech delivery—not as deep experts, but with enough depth to prioritize, problem-solve, and challenge thinking

• Lead end-to-end change management—not delegate implementation to IT

How the best domain leaders build their AI muscle

• Invest 6-10 hours weekly on tech learning (reading, vendor meetings, sprint reviews, conferences, courses)

• Start with business problems worth solving (potential value should generate 20%+ incremental impact)

• Understand that value extraction requires integrating systems—people, data, and technology

• Get hands-on with development teams to build practical judgment

• Develop calibrated judgment on what differentiates top tech talent

What Matters

Three actions for C-suite leaders:

• Put the right people and incentives in place—expect to change 20-30% of current domain leaders

• Launch strategic upskilling programs combining coursework with hands-on practice on real problems

• Fix the operating model: embed engineering talent under domain leaders, shift to persistent funding, align incentives with transformation KPIs

Effortless Tutorial Video Creation with Guidde

Transform your team’s static training materials into dynamic, engaging video guides with Guidde.

Here’s what you’ll love about Guidde:

1️⃣ Easy to Create: Turn PDFs or manuals into stunning video tutorials with a single click.
2️⃣ Easy to Update: Update video content in seconds to keep your training materials relevant.
3️⃣ Easy to Localize: Generate multilingual guides to ensure accessibility for global teams.

Empower your teammates with interactive learning.

And the best part? The browser extension is 100% free.

Check out Guidde

📚 Interesting Reads

• Futurum: Was 2025 Really the Year of Agentic AI?

• OpenRouter & A16Z: State of AI based on 100 trillion token study

• McKinsey: How Boards can Evolve

• Perplexity: Most people use AI agents for Productivity & Learning

• CIO: Agents are poised to rewire the industry and corporate structures

 ➜ Until Next Week

The enterprises winning with AI aren't waiting for perfect models or bulletproof strategies. They're building domain leaders with the muscle to translate technology into transformed operations—and they're doing it now. The gap between leaders and laggards just became a lot easier to measure, and a lot harder to close.

Stay curious,

—The GPTLDR Team

AI, simplified for Decision Makers.