July felt like the month enterprise AI moved from pilots to budgets.
The Adoption Numbers
An IDC study for Microsoft found that generative AI use among businesses jumped from 55% in 2023 to 75% in 2024.¹ The more interesting change is in how it gets used. Teams are moving from a chatbot for everything to tools built for one job.
Banks Hire AI Executives
In June, Danske Bank named its first Chief AI Officer and NatWest named its first Chief AI Research Officer.² Banks are a good early signal. They have decades of structured data, they are used to strict compliance, and they face constant pressure on operating costs. When conservative institutions put AI on the executive team, it has become a core part of the business.
Government Becomes a Customer
On July 14, the Pentagon's Chief Digital and AI Office awarded contracts worth up to $200 million each to Anthropic, Google, OpenAI, and xAI.³ Government deals move slowly, but they tell large enterprises which tools are considered ready for sensitive work.
Capital Keeps Flowing
SoftBank led OpenAI's $40 billion funding round in March.⁴ This month, Mira Murati's Thinking Machines Lab closed a $2 billion seed round at a $12 billion valuation, before shipping a product.⁵
SoftBank's bets line up into a stack: Arm for chip designs, OpenAI for models, and a portfolio of companies building applications on top. It reminds me of early cloud computing. Whoever owns the foundational layer takes a share of everything built on it.
AI Moves Into the Terminal
At the end of June, Google released Gemini CLI, an open-source agent that runs in the terminal and is free for individual developers.⁶ It joins Claude Code and OpenAI's Codex CLI. Developer tools are a leading indicator of platform maturity. Once AI lives in the tools developers already use every day, adopting it stops being a separate decision.
Regulators Build Sandboxes
The UK's Financial Conduct Authority is launching a "Supercharged Sandbox" with NVIDIA, where financial firms can test AI proofs of concept starting in October.⁷ Clear rules and a safe place to experiment remove one of the main reasons regulated companies give for waiting.
How I Prioritize AI Work
- High impact, low risk: process automation with clear ROI, data analysis, developer productivity. Start here.
- High impact, high risk: customer-facing AI and decisions in regulated workflows. Worth doing, with guardrails.
- Low impact, low risk: content generation and chat on top of existing features. Fine, but don't expect much.
- Low impact, high risk: AI for its own sake, or anything without success metrics. Skip it.
Measure What AI Changes
Traditional ROI math counts hours saved, which misses most of the value. I also track how much faster decisions get made, what new work becomes possible, and how quickly the system improves as it sees more data.
What AI work is your team prioritizing, and how are you measuring it?
Sources
¹ Microsoft: IDC's 2024 AI opportunity study
² Danske Bank: new Generative AI leadership appointment and NatWest: Dr Maja Pantic named Chief AI Research Officer
³ CNBC: Anthropic, Google, OpenAI and xAI granted up to $200 million from DoD
⁴ CNBC: OpenAI closes $40 billion funding round
⁵ TechCrunch: Thinking Machines Lab is worth $12B in seed round