The three biggest retail AI funds available to UK investors all own Nvidia. They all own Microsoft. Their top-five holdings lists read like variations on a theme. And yet their annual fees — the OCF, the charge that silently compounds against your return every year — range from 1.00% to 2.86%. The gap is wide enough to demand an explanation.
What the holdings actually show
Start with the facts. Allianz Global Artificial Intelligence, the largest fund here at €8.9bn, charges 2.86% a year. BNP Paribas Disruptive Technology runs $5.6bn at 1.50%. BlueBox Global Technology, the smallest at $2.4bn, charges 1.00%.
All three are retail share classes. There is no institutional-versus-retail distortion here — these are the prices a normal investor on a normal platform would pay.
Now look at the portfolios. Nvidia sits at or near the top of every one of them. Microsoft is second or third in each case. Alphabet and Meta appear in all three top fives. The variance between managers at the highest-conviction positions is essentially noise — a percentage point here, a reordering there.
This is not a coincidence. It reflects how the AI investment thesis has hardened. The companies building and selling AI infrastructure — chips, cloud capacity, the big advertising platforms that monetise AI-generated engagement — are known. The consensus has named them. And when the consensus names the same dozen companies, every manager running an AI mandate ends up in a similar place at the top of the book.
RETAIL SHARE CLASSES · ALL IN BASE CURRENCY
Why it matters for you
There is a version of this story where the fee gap is fair. Active managers would argue — correctly — that the differentiation isn’t in positions one to five. It’s in positions ten to forty. The stock below the headline names, the smaller semiconductor equipment maker or the enterprise software business still two years from monetising AI, is where a manager earns their keep. If Allianz is finding names that BlueBox isn’t, 186 basis points is the cost of that insight.
But here’s the problem with that argument. When the top of the portfolio is identical, the burden of proof on the long tail becomes very high. The manager charging nearly three times as much needs to show differentiation that compounds — not just names you can’t find elsewhere, but names that have worked. Without returns data in this comparison, we can’t settle that debate here. What we can say is that the convergence at the top makes the question harder to avoid.
When every AI fund owns the same top five, the fee gap stops being about conviction — it becomes about everything else.
There is also a scale question worth noting. Allianz Global AI is running €8.9bn. At that size, expressing a differentiated view in mid-cap technology names becomes harder — meaningful position sizes in smaller companies are constrained by liquidity. BlueBox at $2.4bn has more room to move. That could matter. But it’s an argument for the cheaper fund, not the more expensive one.
What to watch next
Three things would sharpen this picture. First, a rotation away from mega-cap AI names — if the market starts rewarding second- and third-tier semiconductor or software businesses over the Nvidia-and-Microsoft consensus, that’s when the long tails of these portfolios are genuinely tested, and fee differences either justify themselves or don’t. Second, fee pressure: as this category matures and more low-cost index alternatives appear, managers at the expensive end — particularly Allianz at 2.86% — may find platform gatekeepers asking sharper questions. And third, the next round of full portfolio disclosures. The top five is a summary; the full holdings list is where genuine differentiation either exists or doesn’t. That’s the document worth reading.
Past performance is not a guide to future returns. Capital is at risk.
Funds Benchmark provides research and tooling for institutional and private investors. Nothing in this note is investment advice or a recommendation to buy or sell any specific fund. Past performance is not a reliable indicator of future results.