The index is risky. Why Antipodes is finding value beyond the AI trade

Most investors go looking for reasons an investment will work. Vihari Ross spends a lot of time thinking about how it could go wrong.

That might be the actuary in her. Before becoming a fund manager, Ross trained in actuarial studies, where probabilities, distributions and unlikely but painful outcomes are part of the furniture. Today, as a Portfolio Manager at Antipodes responsible for reviewing investment cases, those habits have shaped how she thinks about stocks.

Her favourite question is simple: “What’s the distribution curve?”

If the possible outcomes are reasonably narrow, there may be grounds for conviction. If every outcome looks equally probable, the answer is more straightforward.

“It means you’ve actually got no idea. Don’t buy the stock.”

That focus on risk runs through Antipodes’ approach to value investing. But it does not mean restricting the portfolio to conventionally cheap shares. Indeed, Ross argues some of the best opportunities today sit in parts of the market investors would normally associate with growth.

The global index has become highly concentrated around a handful of AI-related companies, while valuations beneath the surface have become increasingly dispersed.

Ross thinks the index looks risky, even as individual opportunities remain plentiful.

In this episode of The Rules of Investing, we discuss what value investing looks like in 2026, why market concentration has reached levels seen only a handful of times in history, and whether investors have been too quick to sort companies into AI winners and losers.

We also explore the investment lessons Ross has carried through a 25-year career, and the US senior housing company she would be comfortable owning if markets closed for the next five years.

Start with what can go wrong

Downside protection is not just about avoiding unpleasant numbers on a screen. The arithmetic matters.

Lose 50% and you need to make 100% simply to get back to where you started. Lose 25%, and the recovery required is far less punishing.

That asymmetry is one reason Antipodes uses what Ross describes as a “pre-mortem” when assessing investment ideas. Before buying, the team asks what could break the thesis, which assumptions matter most, and how severe the consequences would be if they are wrong.

The aim is not to eliminate uncertainty. Markets have stubbornly declined to provide that service. It is to distinguish between uncertainty you can be paid to take and outcomes capable of overwhelming the original investment case.

Downside risk also influences how Antipodes builds the portfolio. An attractive upside case is not enough if the downside is catastrophic, while a stock with a less spectacular upside can still be compelling if the range of outcomes is unusually favourable.

The process is reinforced by Antipodes’ team structure. Sector teams build their own model portfolios, while investment cases are challenged by people with different backgrounds and biases. Ross sees disagreement as useful rather than inconvenient – disagreement can expose assumptions that might otherwise survive unchallenged.

Value doesn't have to look cheap

Antipodes applies the same scrutiny to what counts as ‘value’. A traditional value screen might simply identify the cheapest quintile of the market. Ross thinks that risks confusing a low valuation with an attractive investment.

“Some of those companies are cheap for a reason.”

A beaten-up incumbent facing a temporary cyclical downturn can be interesting. A company being permanently disrupted is something else entirely.

The reverse can also be true. A business trading on a relatively high multiple may still offer value if its growth and resilience are being underestimated.

That is what Antipodes means by “pragmatic value”: judging price relative to the durability and growth of the underlying business rather than demanding that every holding conform to a particular PE multiple.

Microsoft provides a useful historical example. When Antipodes was founded in 2015, Ross says the company traded on roughly 12 times earnings and was treated as something of a has-been technology business. The initial opportunity looked like conventional mean reversion.

But as Azure and cloud computing transformed Microsoft's growth profile, the investment case changed with it. Even after the original valuation discount closed, the company could remain inexpensive relative to what its future earnings had become.

That discipline cuts both ways. Ross points to PayPal during the 2022 growth sell-off as an example of a company where a falling share price did not automatically create value. Its valuation had embedded far stronger growth than the business eventually delivered.

A low price is only useful in context: investors still need to understand the quality, resilience and growth they are buying.

The index is expensive. Underneath it, things get interesting

The current market gives Ross plenty of examples. She estimates the “AI Big 10” – the Magnificent Seven plus Broadcom, AMD and Micron – now represent around 40% of the US benchmark. Broaden the lens to include software, internet businesses and the industries caught in AI’s gravitational pull, and more than half the market has some connection to the same theme.

The valuation picture is less tidy. Some AI-related companies remain attractive, while parts of the market outside the theme are expensive. What stands out is the spread between the extremes.

Ross describes the US market and hardware sector as expensive relative to their own histories. At the same time, smaller and mid-sized companies, Europe, emerging markets and parts of software trade at discounts.

Even the US market looks very different once the mega-caps are stripped out.

That is why she is wary of describing a passive global-equity allocation as neutral.

“There’s nothing passive about a passive allocation.”

Today's benchmark embeds large bets on the US, mega-cap companies and AI, whether an investor consciously chose them or not.

History offers some uncomfortable precedents. The current level of concentration is comparable with previous peaks around the railroads, the Nifty Fifty, Japan in the late 1980s and the technology bubble.

History offers no useful clock for when concentration will reverse. It does, however, show that leadership can broaden sharply after a peak.

Ross points to 2000 as an example. While the headline index struggled as the technology bubble unwound, the median stock actually rose.

“There’s risk here at the index level,” she says, “but there’s opportunity more broadly.”

Not every AI 'loser' is losing

Technology provides some of the clearest examples. Markets have spent much of the past few years rapidly sorting companies into AI winners and losers. Ross thinks some of those judgements have been far too hasty.

Memory stocks show the opposite problem. Supply shortages have driven a powerful earnings cycle for companies such as Micron, SK Hynix and Samsung. But memory remains a commodity, and commodities have an inconvenient tendency to respond to high prices with additional supply.

Ross argues investors were simultaneously pricing peak margins and elevated valuations while paying too little attention to the industry's long history of cycles.

Software has faced almost the opposite treatment. Broad fears of AI disruption have pushed down valuations across businesses with very different economics and competitive positions.

Booking Holdings is one example Antipodes bought during the sell-off. Despite often being grouped with software and internet stocks, its competitive advantage rests heavily on the enormous inventory of accommodation it has assembled and the relationship it has with smaller hotels.

An AI agent might change how consumers search for a holiday. It does not magically recreate that inventory.

Instacart presents a similar case. Beneath the software interface sits a network connecting supermarkets, customers and physical logistics.

As Ross puts it:

“That’s not going to be vibe coded away.”

The picture is different again for enterprise software. Companies such as Microsoft and Salesforce already sit deep inside corporate workflows, hold sensitive data and operate within security and regulatory structures that are not easily replaced by somebody building an app with an LLM over lunch.

Some incumbents will undoubtedly be disrupted. Others may become the very distribution channels through which AI reaches enterprise customers.

The challenge is working out which is which before the market does.

Ross’s process starts with accepting that uncertainty cannot be removed. The job is to understand the range of outcomes, what the market has already priced in, and whether the potential return compensates for the risk of being wrong.

That process can lead Antipodes to an unfashionable cyclical or a technology company trading on a high multiple. But if the distribution of outcomes is too wide to form a sensible view, Ross’s rule is simpler: ‘Don’t buy the stock.’

Learn more

Antipodes global equity strategies give investors access to high-conviction, actively managed portfolios focused on leading businesses across developed and emerging markets. You can find out more by visit their website.