There's something crazy happening in the world of investing as AI is soaking up all the money available, even your pension fund. (Photo: Bruno Aguirre / Unsplash)For more than four decades (1980-2020), retirement funds operated in an environment cushioned by falling inflation, declining bond yields, cheap capital and expanding globalisation.
As Mario Fisher, chief investment officer at Momentum Systematics, told the Institute of Retirement Funds Africa (Irfa) 2026 Conference at the CTICC, those tailwinds have vanished.
“Many investment frameworks were built for a world that no longer exists,” he said. “We are not experiencing a normal cycle. We are experiencing a regime change at the moment.”
Fisher reckons that efficiency is being replaced by resilience and security. Geopolitics, fiscal deficits, supply chain duplication and carbon transition expenditure are driving higher structural inflation and interest rate volatility. In this environment, traditional 60/40 equity-bond portfolios and sovereign bonds no longer offer automatic diversification.
Beyond compliance
Irfa chairperson Nancy Andrews took the stage directly after him in a one-two punch that knocked the puff out of the chest of many fund managers.
She made a different point about retirement fund trustees who can no longer evaluate governance purely by procedural compliance; success is measured by actual post-retirement purchasing power.
“Governance is something that we learn daily. Members don’t experience our governance structures; they experience the outcomes. If we do not get it right, they feel the brunt of it... Strong governance does not start with having the right answers, but with asking the right questions.”
This focus on member experience is echoed by financial regulators. The Financial Sector Conduct Authority (FSCA) has signalled a decisive transition from compliance-based supervision to outcomes-based regulation, focusing heavily on value for money, cost transparency and active ownership.
But these original peddlers of patient capital must be wary of the AI bubble. While technological breakthroughs offer immense long-term potential, the current market fervour surrounding AI presents acute risks for institutional capital.
The biggest challenge of the AI investment bubble is the stark mismatch between institutional liability horizons (10-30 years) and the unproven long-term commercial mechanics of AI technology.
Executive chairman of Gaia Capital, Mich Nieuwoudt, has already shared controversial takes about infrastructure investment with Daily Maverick, but even he is cautious with this particular beast.
“I am paid, and we make a lot of effort to think about how the 10-plus-year future is going to look on AI and data centres,” he explains. “If anyone tells you they know how it’s gonna look five years out there, they are liars...”
“I don’t know how to invest pension fund money in something that I have no clue how the mechanics of it is going to work in five years’ time. I am 100% certain... that in 20 years’ time we’ll still be using electricity. I am unsure whether in 20 years we’ll still be using AI as we are using it today.”
Institutional mismatch
A critical real-world bottleneck created by the AI surge is its immense demand for power generation and data centre infrastructure, and Nieuwoudt points out that underlying AI investments are often mispriced because investors fail to account for long-term power contracting risks.
“They are priced as electricity, but under that is you are selling the electricity at a specific rate for a specific period into the future and you don’t know what that is going to be over a 10-year period...” he says.
“In AI, I think everybody can agree that no one knows, or there’s a high spread in what will be and what people think they would be... Just the fact that there’s a wide variety of opinions out there should give you a risk premium that you must price into your calculations.”
When institutional portfolios over-concentrate in speculative technology assets without requiring an adequate risk premium, they expose members to severe downside shocks when expected outcomes diverge from reality.
But rather than treating AI as a blanket growth driver, institutional investors must analyse its stratified value chain, distinguishing between hardware providers, infrastructure operators, model developers and corporate adopters.
“AI represents a potential transformational investment theme... but the challenge for investors is distinguishing between the providers, the adopters and the losers,” Fisher says, warning of the earnings dispersion.
“Importantly, AI is increasing potential for earnings dispersion. You could have companies in the same sector with huge winners and huge losses. But if you pick the sector, your net effect is just zero.”
A reverse infrastructure trap
What has become apparent is that a fundamental flaw in traditional retirement fund asset allocation is holding national economic infrastructure indirectly through government debt rather than direct equity ownership.
Despite South African regulatory frameworks (such as Regulation 28) permitting up to 45% allocation to infrastructure, institutional allocations remain in low single digits.
Nieuwoudt chalks this up to chronic under-allocation to institutional inertia and knowledge gaps.
“Who makes those decisions? It’s two people: the asset consultants and the trustees. But no one is paid to learn a new asset class.”
He points to emerging market peers like Chile – where local pension funds own roughly half of the country’s domestic infrastructure – African funds remain overly dependent on Development Finance Institutions (DFIs) and international debt pricing.
Nieuwoudt’s point about retirement funds being ploughed back into the economy to grow jobs and build the critical infrastructure needed to kickstart the slowing economy is important – just that he wasn’t on stage at Irfa. DM
Comments
Scroll down to load comments...