Ninety One Grew 19% and Still Lost £4.2 Billion — the EM Trade Is Right, Just Early
Interim results show a fatter margin, a 5% dividend lift and one number management admits it did not budget for: the outflow
N91.L · Earnings Call · 2026-11-16
The number nobody budgeted for
Ninety One's half-year to 30 September 2026 has all the surface markers of a good print. Assets under management rose 19% to £181.4bn (helped enormously by the Sanlam consolidation), average AUM jumped from £139.7bn to £179.6bn, management fees climbed 21% to £351m, adjusted operating profit rose 19% to £117.8m, the operating margin widened to 32.5% from 32.1%, the compensation ratio fell to 42.3% from 43.6%, and adjusted EPS rose 5% to 8.8p with the interim dividend lifted in lockstep to 6.3p.
And then there is the other number, the one the CEO led with: “Assets under management grew by 19% to GBP 181.4 billion. We experienced net outflows of GBP 4.2 billion. This was not anticipated at year-end.” — Hendrik du Toit, CEO · 2026-11-16
That is a rare thing in an asset-management interim: an explicit admission that guidance was wrong. Analysts had already taken the red pen to it, and Hendrik du Toit did not hide. Asked about the second half, he answered with the least promotional sentence in the transcript: “I'm so hurt after the last signal I gave you that flows would be good. I mean we were budgeting for pretty big inflows. I didn't think I was just hoping.” — Hendrik du Toit, CEO · 2026-11-16
But I think it could be better than this was not a good experience... very rarely do we have 3 things or 2 categories underperforming at the same time. So let's see. But what we do know is you don't switch the inflow on immediately once you recover your numbers, you've got to convince clients that you've actually -- you're actually sustainable again.
The net outflows have two distinct causes, and only one of them is a management problem. The first is investment underperformance in two places: the quality style in global and international equity, and South African domestic equity. The second is a funding-timing problem — performance is fine in emerging markets and commodities and resources, but the money simply did not land. There is a won-but-unfunded pipeline above £2.5bn, skewed to the UK, part of which may fund after the year-end.
The thesis is right, the tape is wrong
The strategic bet at Ninety One is unchanged and, on its own numbers, unchanged for good reason. Over three years the CEO argues emerging-market hard and local currency bonds and EM investment-grade credit have delivered double to three times the outcomes of developed-market peers, even as equity investors remain fixated on US growth. Emerging markets is, unsurprisingly, the single highest-momentum theme in the company's own keyword set for the quarter, and the Sanlam deal was explicitly framed as a deliberate re-weighting toward EM and fixed income at a point when the equity bull market looks mature.
What is interesting is *why* the rotation is late — and here the global tape corroborates management's diagnosis rather than contradicting it. The 30-day market decliners are an AI-infrastructure roll-call: PCIe Gen 6 names off roughly 29%, high-performance computing and data-center complexes down double digits, cloud service provider baskets soft. That is precisely the "AI trade unwinding" du Toit points to when he explains why the Hong Kong IPO window matters to his Asia joint venture: “what we've seen in Hong Kong lately is the IPO window is largely closed for the moment. So I think that's into the new year.” — Hendrik du Toit, CEO · 2026-11-16 The trade that has been starving the EM rotation is now itself deflating — but the rotation still has not been funded. Being early is indistinguishable from being wrong until the flows arrive.
The quality style explanation is the most quotable line in the call: “the quality style will remain in a tough place while markets remain obsessed with growth stocks.” — Hendrik du Toit, CEO · 2026-11-16 Management's answer is to support the team rather than pressurise it. Whether clients share that patience is the open question.
The fee arithmetic is where the real pressure sits
The average fee rate fell to 39 basis points from 41.5, of which only about one basis point was organic — the rest is the Sanlam book, a £23–24bn asset pool carrying a lower rate. Management kept its 38–40bp guidance and continues to imply 1–1.5bp of annual fee pressure, but du Toit moved the argument deliberately: “I'd worry less about fee, more about alpha and operating margin.” — Hendrik du Toit, CEO · 2026-11-16 That is the correct framing for an active manager, and also a warning that the 32.5% margin is being defended by scale and technology rather than price.
Performance fees fell to £12.1m and management was blunt about the near term — "I wouldn't get excited. It's GBP 1 million here and there." The real pool sits in Ninety One Asia's crossover vehicle and in private markets, where the company has just exited European credit opportunities in favour of a partnership with Sound Point, is close to seeding an EM debt hedge fund, and expects first closes on two infrastructure vehicles in the fourth quarter. Private-credit asset raising "takes time", and the call said so repeatedly.
The AI slide investors will skim past
Buried in the middle of the presentation is the most forward-looking disclosure of the half: self-scored AI adoption at 9 out of 10, experimentation 9 out of 10, data 5 out of 10, transformation 3 out of 10. The adoption detail is striking: “In September, about 90% of our people were active users of Claude. This figure compares with 50% 6 months earlier.” — Hendrik du Toit, CEO · 2026-11-16 A quarter of those users now work in Claude Code — agents changing how tasks are actually done on desk, not just chatbots answering questions. That places a mid-cap London asset manager inside the same personal AI agent wave the global keyword set is tracking, and doing it faster than most of its peer group.
The honest part is the cost. Business expenses rose 22% to £91.2m with technology the largest driver, and the CFO explicitly said she is not seeing a decrease in cost in the second half. Du Toit's rationale is candid: “we should have done that 7, 10 years ago when it was really cheap. So that costs now and you burn a lot of tokens and that to kind of change the model.” — Hendrik du Toit, CEO · 2026-11-16 The AI enabled rebuild is a margin cost today against an unproven productivity gain tomorrow — and management declined to claim the gain.
Why it matters
Strip out the Sanlam optics and Ninety One is a capital-light business earning a 32.5% margin, covering its dividend 220% with £147.1m of surplus capital, and having returned over 70% of its IPO market capitalisation through buybacks and dividends since listing. The buyback continued — £25m in the period, 59.3m shares retired cumulatively. None of that is in question.
The story is the gap between a structurally correct thesis and a market that has not yet paid for it. A 19% AUM gain, a 5% dividend increase and an expanding margin, against a £4.2bn outflow that management had told investors would not happen — and then a refusal to promise the second half, ending with "we look forward to seeing you at the end of the year with better news." The H2 print is now the whole investment case.