The massive AI/hyperscaler capex boom is a sound investment that will eventually deliver a real return, rather than being a bubble that fails to pay off.
“my view is that it's real, that I think there will be a return on all this capex ... eventually there will be a return on that investment”
View on YouTube
Explanation
This is an explicitly long-horizon claim ('eventually there will be a return on that investment') about the multi-year AI capex cycle. Only days have passed since this episode aired (published July 31, 2026); there isn't yet a meaningful multi-year track record to judge the capex-to-return thesis against.
Because elevated Treasury yields (driven by persistent federal deficit spending) make bonds more attractive than richly-valued AI/tech stocks, more asset-price bubbles in AI-related equities will pop unless US fiscal and monetary policy changes course.
“we are going to end up seeing more bubbles pop and more of these assets ... that we've kind of inflated”
View on YouTube
Explanation
Whether elevated Treasury yields trigger further AI/tech asset bubbles popping is an ongoing, multi-month-to-multi-year macro dynamic. Not enough time has passed since this prediction (made in an episode published July 31, 2026) to assess it either way.
New AI efficiency techniques will soon be demonstrated that cut token consumption by roughly 50-75% for the same task without loss of capability.
“there is some incredible efficiencies that I think are about to be demonstrated which effectively cut token consumption by about 50 to 75% for the same task”
View on YouTube
Explanation
New AI efficiency techniques cutting token consumption by roughly the predicted magnitude have indeed been demonstrated: 2026 approaches like prompt caching, semantic caching, and 'code mode' for tool use are documented delivering 50-80%+ reductions in token/compute costs for comparable tasks, matching the prediction closely and on a similarly short timeline ('about to be demonstrated').
Solar (with batteries) will become so cheap and dominant in power generation that small modular nuclear reactors (SMRs) will be economically obsolete by the time they reach production.
“by the time any of these SMRs actually get near production, the TCO of solar will be like 10 or 12 per megawatt hour and it will be 80% of all the power generation”
View on YouTube
Explanation
This prediction is explicitly conditioned on a multi-year timeline ('by the time any of these SMRs actually get near production'); SMR production timelines extend years into the future, so there is nothing yet to score this against.