Who actually gets ROI from AI.
A global KPMG survey puts numbers on which firms get value from AI: the ones where the CEO owns it, with real visibility into cost and return. For a fund, AI value tracks governance, not how many tools you buy.
Hi folks,
Last week a regulator switched off a frontier model. This week a global survey put hard numbers on which firms are actually getting their money back from AI.
The one thing
The firms getting ROI from AI are run differently, not tooled differently
- KPMG's Q2 global AI survey (2,145+ senior leaders across 20 countries) found firms where the CEO or an executive owns AI decisions are well ahead: 57% report meaningful value versus 21%, and 14% have established ROI versus 4% (KPMG).
- Firms with real cost visibility were five times more likely to have established ROI (15% versus 3%). Scaling, not spend, was the top barrier at 65%.
What this means in plain terms:
- For a fund, AI value tracks governance, not licences: a named owner at the top, and visibility into what it costs and returns.
- The "give everyone a chatbot" approach is the one the data says does not pay. Pick a process that runs across the team and rebuild it, with an owner.
In the mix
- Anthropic put an AI teammate inside Slack (Anthropic)
- Claude Tag joins a channel, picks up tasks when mentioned, learns the organisation over time and runs multi-step work. Tools and permissions are set per channel.
- Anthropic says it already writes 65% of one team's code.
- Why it matters: the first agent genuinely usable by a non-technical team, with a governance surface (per-channel permissions, an audit trail) a compliance-minded fund can actually reason about.
- Google lost two senior AI researchers in a week (TechCrunch)
- A Nobel laureate left DeepMind for Anthropic; a co-author of the Transformer paper left for OpenAI. Alphabet shares fell about 5%.
- Why it matters: the labs your tooling depends on are not interchangeable, and their relative standing is moving quarter to quarter. Worth tracking if you are betting on one provider.
From my week
- Making AI legible for teams
- This week, we built plain-English AI-literacy material for finance teams: how a model actually works, what it can see, and how you instruct it.
- We also addressed the governance questions that impede fund adoption (data residency, client-data confidentiality).
- Wrote up the first month
- I published "Stop Thinking Small About AI": most organisations use AI to make individuals faster when the prize is rebuilding processes so the firm compounds context. Execution is mostly solved, context is the work (Medium).
Levercon is the AI team for investment funds. We embed alongside your team to automate workflows, build custom tools, and compound operational gains. Ultimately we help Australian based investment funds leverage AI.
Book a strategy call.
See where AI fits in your fund.
We'll show you where AI lifts efficiency and performance across your organisation.