Menlo Ventures' new LLM report is pure gold.

150 enterprise leaders surveyed.

$8.4B in AI spending tracked.

Here's what you need to know:

📊 The Market Shift Nobody Expected

Anthropic: 32% market share

OpenAI: 25% (down from 50%)

Google: 20% and climbing

Meta's Llama: 9%

DeepSeek: 1%

Claude Sonnet 3.5 changed everything.

💰 Follow The Money

LLM API spend more than doubled in 6 months.

From $3.5B → $8.4B.

74% of workloads = production inference now.

We're done experimenting. We're shipping.

🔥 Code Generation Is The Killer App

42% of devs choose Claude for coding.

OpenAI only captures 21%.

$1.9B ecosystem built in one year.

This is why Cursor, Windsurf, and Bolt exist.

📉 Open Source Reality Check

Enterprise usage dropped from 19% to 13%.

Performance gap: 9-12 months behind closed-source.

Chinese models dominate but enterprises won't touch them.

🎯 What Actually Drives Decisions

66% upgrade within same provider.

Only 11% switch vendors.

Performance beats price every single time.

Despite 10x price drops, everyone pays for frontier.

The Big Insight:

Anthropic won by training models as agents.

Not chatbots that answer questions.

But systems that think step-by-step and use tools.

2025 = Year of agents.

And Anthropic built for it first.

Which stat shocked you most?

♻️ Repost to share what the AI hype accounts won't tell you

Full report👇