McKinsey new report on AI - #36

Series “Laboratory”

Introduction

Welcome back to Laboratory, our weekly deep-dive tracking the forces that are redefining AI. Two weeks ago we mapped the supercomputer arms race; today we pivot from hardware to code—from private data centers to the open repositories anyone can clone.

A new McKinsey–Mozilla–McGovern survey of more than 700 tech leaders signals an inflection point: open-source AI has become a strategic pillar for cutting costs, accelerating innovation and attracting talent. Yet it also exposes fresh fault lines—security, compliance and IP risk—that will separate tomorrow’s winners from the rest.

Here’s what we’ll unpack:

• Economics of Openness – Why most firms now say community models beat proprietary ones on build-and-run cost.

• Performance & Adoption – Evidence that satisfaction with open models outpaces dissatisfaction ten-to-one, and how usage is spreading across the stack.

• Guardrails that Matter – The emerging tool-chain—prompt filters, Llama Guard, sandboxed runtimes—that lets enterprises go open without tripping compliance wires.

Join me as we explore why openness is winning, where the pitfalls lie, and how the smartest players are balancing freedom with discipline to build the next generation of AI products.