LPPI: Keeping AI risk in check

As investor unease increases with the growth of AI’s portion of the capital markets and tech companies’ huge spend on AI and data centre infrastructure, measuring and managing exposure to the technology has become a priority.

At the UK’s £59 billion ($79 billion) LPPI, which manages assets on behalf of nine Local Government Pension Schemes, exposure to AI crops up across the portfolio in different asset classes. Like many investors currently, LPPI has meaningful exposure to AI in aggregate, LPPI’s chief investment officer Richard Tomlinson tells Top1000funds.com.

For example, LPPI has around 10 per cent direct exposure to hyperscalers in the reference benchmark in its main public equity fund. The pool offers partner funds exposure to a range of equity products, the largest of which is the £11 billion actively managed LPPI global equities fund which invests through a combination of internally managed and third-party managed investments.

“By the time you put in Meta, Amazon and Google in the MSCI ACWI index, exposure to AI and technology becomes meaningful,” he says.

But exposure extends beyond just the tech giants.

“We, and others, have looked at how much of the broad equity market is linked to the AI trade, and external analysis suggests that a significant share of recent returns has been associated with AI-related companies. If you look through MSCI World and think how much AI exposure there is in there, it’s large because of all the ancillary business and earnings that are being driven by AI.”

LPPI, was one of the smallest Local Government Pension Scheme pools, but emerged as one of the big winners in the UK government’s drive to reduce the number of LGPS pools which collectively manage £392 billion ($522 billion) last year.

Drawing on inspiration from the Canadian model, the former Chancellor Rachel Reeves reduced the number of pools from eight to six in a bid to boost efficiency and scale, and enable big-ticket investment in the UK economy.

Managing the risk

The team takes a quality approach to its tech exposure in the equity allocation, seeking to avoid speculative stocks and trying to direct capital into companies that will compound quality over time.

LPPI comfortably owns hyperscalers involved in “genuine corporate activity” but has so far steered away from so-called neo-cloud business models. These businesses are the new generation of cloud providers which have emerged to serve AI workloads specifically, rather than offer the broad, general-purpose cloud platforms of the traditional hyperscalers.

“Hyperscalers are more diversified cloud platforms compared to subtly different businesses like CoreWeave, a specialist AI infrastructure provider,” he explains.

In another approach, LPPI aims to select companies that will perform in a broad range of future scenarios and find exposures they believe have a positive upside and a more resilient downside, depending on how AI evolves.

“We try and find businesses that can play AI and the demand for data, whichever way it goes. We want to be in a good position to benefit, and have exposure with more resilient downside characteristics, as opposed to having a pure play on one thing happening.”

It means the investor avoids equity exposure to the higher beta components that are volatile relative to the wider market.

“We seek to avoid allocations that work well in a compute boom, but may be more exposed if the AI investment cycle develops differently,” he says.

Underwriting the risk requires looking at the asset not in the context of AI but broad demand for future cloud and data infrastructure, he continues.

“Our underwriting philosophy does not bake in an exponential growth via AI and compute. We have deliberately not taken a single bet on a straight line up in AI,” he says, avoiding an approach where AI stocks go up smoothly and indefinitely.

Tomlinson is broadly convinced that AI will ultimately be transformational as a general-purpose technology, but remains uncertain how the return on investment will perform in the next five to 10 years.

“How we get there will be bumpy,” he says.

As well as equity exposure via hyperscalers and companies in the data-centre supply chain, LPPI also has direct exposure to data centres. This comes through the infrastructure allocation, including pooled funds and co-investments, and potentially through real estate.

Although data centre exposure in real estate is still limited, he notes LPPI’s managers are increasingly looking at opportunities that re-use industrial land with access to power.

fit for the future

Tomlinson explains that LPPI’s investment approach sits beneath partner funds’ strategic asset allocation decisions and above individual mandate selection. It is focused on translating partner fund objectives into practical portfolio construction and implementation. It means the pool implements client SAAs, but also thinks on a portfolio-wide basis to align the overall portfolio as best as possible with client objectives.

Fit for the Future, the UK government’s reform agenda for the LGPS, includes the requirement for pools to provide investment advice.

LPPI has always provided principal investment advice tailored to partner funds’ unique funding levels, liability profiles and risk appetites, explains Tomlinson.

“Most of my grey hair comes from dealing with the total portfolio because we have been doing this for 10 years already and it’s complex and thorny and has no easy answers,” he says.

“In the past, pools said to their partner funds ‘you choose what split you want between equity fund A or B.’ That’s changed, and now pools have to make that determination on behalf of their partner funds and ensure it aligns with the agreed mandate.”

Although LPPI is well versed in the advisory piece, other elements of Fit for the Future have required real change.

“We have gone from a high-touch CIO model to one where we have hired client managers to scale our architecture as we have doubled AUM and tripled client numbers this year. It involves a level of tooling, process building and governance that is both positive and challenging.”

Pools are required to offer nine asset classes under FfF. LPPI is building a structure that defines the asset classes on offer and the different strategies within those asset classes. From this, the team can create strategy mandates that define what the strategy is and what it does.

“From these building blocks, we can assemble a portfolio and implement it on behalf of our partner funds,” he concludes.

“We are supporting partners to articulate target return and risk appetite in a more formal and structured way. It cuts across the structural design of the investment process and architecture, advice and guidance, and the softer pieces around working with and supporting new clients. On top of all this, we are also managing the portfolios and integrating legacy assets.”