Katrina Dudley, Senior Investment Strategist at Franklin Templeton, is not backing down from her AI infrastructure thesis. In a Bloomberg appearance on June 11, Dudley argued that the current wave of AI capital expenditure is still in its early innings, describing it as a “decade-long cycle” that has barely gotten started.
That framing matters. Wall Street’s mood has shifted noticeably toward demanding proof that all those billions flowing into AI infrastructure will actually generate profits. Dudley’s counter-argument is essentially: relax, this is what every major infrastructure boom looks like in year two.
Samsung’s wild ride and the overcapacity question
Part of the conversation centered on Samsung Electronics, whose shares have surged roughly 169% year-to-date as of late June. The Korean chipmaker has been a primary beneficiary of insatiable demand for high-bandwidth memory products, which are critical components in AI training and inference hardware.
But the stock has recently slumped on overcapacity fears. The concern is straightforward: when every major chipmaker races to expand capacity simultaneously, supply eventually overshoots demand, and margins get crushed.
