Markets in 2026 are a study in extremes. The AI-driven euphoria has lifted a narrow cohort of winners to unimagined highs, while a long tail of former tech market darlings have halved in value. For value-disciplined investors, this divergence is the opportunity set.

A recent conversation with Caroline Cai of Pzena Investment Management, a global value house we hold across many portfolios, neatly captured the current moment. Pzena bought Samsung when nobody wanted it. The thesis worked. Memory pricing has gone vertical, gross margins are tracking from a long-run 35–40% toward 80% this year, and as Caroline pointed out, the maths is more dramatic than it looks: doubling gross margin from a fixed cost base sextuples profit. Earnings are up by leaps and bounds.

And yet they have been selling.

Pzena anchors on mid-cycle earnings, not the peak. Memory is a commodity business in the most classic sense — temporary shortages, price spikes, and then the inevitable demand response as price-sensitive end markets (handsets, PCs, industrial users) simply use less. Add China's well-documented ability to close supply gaps quickly, and the through-cycle picture looks very different from the screen-printed forward P/E of six or seven times. As Caroline put it, quoting Jurassic Park: "life finds a way." In commodity industries, supply always does.

Far more interesting than what they are selling is what they are buying.

The same market that has bid up the AI beneficiaries has punished anything perceived to be AI-displaceable. IT services names like Cognizant, payments businesses like Global Payments, these are the new fallen angels. The bear narrative is that vibe-coding will dissolve the need for systems integrators. Pzena's counter is that no large enterprise, an ANZ, a Nike, actually migrates to a new platform without the technologists who have always done that work. Current revenue softness reflects cyclical pressure on customer IT budgets (tariffs, cost-of-living, deferred discretionary spend), not structural obsolescence. Crucially, bookings and backlogs continue to grow. In some cases, the existing backlog alone justifies today's market cap, with new bookings essentially free optionality.

Closer to home, CSL has just appeared on Pzena's screen for the first time. A business that for years traded as a quality compounder is now flagged by an emotionless valuation model as deeply attractive. The work now begins: is something genuinely broken, or is this a cyclical air-pocket with adequate downside protection at today's price?

This is where polarised markets earn their keep for active investors. When the index is dominated by a handful of names priced for perfection, the opportunity set widens dramatically among the unloved. We continue to underweight expensive US mega-caps to some extent and lean into emerging markets and global value strategies for precisely this reason. Not because the consensus winners are wrong, but because the margin of safety, today, sits with the businesses everyone else has given up on.

There is always a grain of truth in the bear case. The question is whether the price already pays for it.