The four largest hyperscalers are expected to spend roughly $700 billion on AI capital expenditures this year, a scale with few historical parallels outside the railroads. Equity analyst Irena Petkovic examines what that spending is actually buying and what has to be true for it to pay off. She explains how the data center has shifted from a cost center to a revenue-producing asset, walks through the four ways hyperscalers monetize compute, and lays out the token economy that underpins the newest business model. She also weighs the early evidence of returns against the risks around token pricing, debt financing, and public backlash, and describes how the team positions the portfolio around all of it.

Labels and mental shortcuts help us navigate a world of too much information, but they turn dangerous when they harden into rigid identities that dictate decisions even as facts change. The article argues for "integrative thinking", a term coined by Roger Martin: holding opposing ideas in tension to find a third path rather than collapsing into either/or choices. For investors, that means keeping the timeless questions of "quality" while staying forward-looking and agile enough to reassess as the world shifts—embracing the "and."

Memory prices have surged, and no one can say where they will be in 2029 or 2030. Equity analyst Shan Rui Yeo explains how the team values SK Hynix when the range of outcomes is that wide: a Monte Carlo framework that runs a discounted cash flow through thousands of scenarios and prices the business on a distribution of returns rather than a single forecast. He also unpacks the leveraged single-stock ETFs that have been amplifying the stock’s daily swings, and why flow-driven volatility can be a gift for long-term investors. Above all, he makes the case that memory can be cyclical, structurally growing, and wealth-creating all at once.