Back Research Notes AI “Bubble” or $Trillion Shift? Published on September 28, 2025 By Jordi Visser In this week’s video , the “AI is a speculative bubble” fears have risen once again. This time I go through Jensen Huang’s recent interview to highlight why this is the beginning of a trillion-dollar global structural shift. I cover how Tesla’s soon-to-be-released FSD v14 could accelerate investors focus on humanoids, why compute and power shortages are defining this cycle, and how the labor market is breaking away from GDP as AI agents begin to displace white-collar work. Why this is an important input for trading decisions in 2026. Commodities like copper and DRAM are flashing stress signals, while China’s massive CapEx surge shows this is not a US bubble but a global race to build capacity and it is far from slowing. The conclusion: this is not dot-com 2.0. Scarcity in compute, energy, and labor adaptation means this demand shock is structural. AI is accelerating faster every day. The opportunity lies in identifying the choke points, semiconductors, power grids, humanoids, and policy responses. The AI transformation is happening faster than expected, and positioning ahead of the curve will be critical. Timestamps 00:00 – Theme • The real risk isn’t AI itself, but losing to competitors who adopt it first. • Sets the stage for semiconductors, AI agents, and labor displacement as the drivers of this cycle. 01:37 – GDP & Inflation • U.S. growth reaccelerates: Q3 tracking near 4% despite tariff headwinds. • Core PCE continues to remain muted despite tariff impact, with markets still pricing in Fed rate cuts for October and December. 03:13 – Commodities & Energy • Copper’s multi-year bull cycle, linked to AI infrastructure demand, gains momentum. • Power grid tightness (PJM states) highlights rising retail costs and strained capacity. Energy stocks begin to break out. 06:44 – Jobs & AI Disruption • From Bosch’s layoffs to Fiverr’s restructuring, firms admit AI is replacing even low-cost human labor. • A structural break: labor markets no longer track GDP growth, pointing to long-term policy stress. 11:35 – Mercor & RLHF • Startup trains models with domain experts (law, banking, healthcare) to accelerate specialization. • Highlights how capitalism is “training its own replacement,” monetizing human expertise for model obsolescence. 16:19 – GDP Measurement • Traditional GDP relies on labor income feedback loops, but AI abundance undermines that foundation. • As scarcity fades, GDP fails to capture real value creation forcing a rethink of economic measurement. 19:21 – AI Bubble Debate • Bain’s $800B “shortfall” forecast overlooks new revenue streams like robotaxis, AI healthcare, and agents. • Historic parallels: 1990s internet and 2007 smartphones also underestimated ecosystem growth. 28:41 – Jensen Huang Takeaways • Inference demand projected to scale by 100–1000×, driving exponential compute needs. • Every new chip generation delivers higher tokens-per-watt efficiency, reinforcing Nvidia’s moat. 35:08 – China’s CapEx Surge • Alibaba’s pledge to 10× data center capacity shows hyperscalers see no capex slowdown. • $625B+ in clean energy spend (10% of GDP) positions China to leapfrog U.S. in AI infrastructure. 38:03 – DRAM & Semiconductors • Tight DRAM supply benefits Samsung, SK Hynix, and Micron, while bottlenecks ripple downstream. • Second-order beneficiaries like Teradyne and ASML see demand pulled forward as testing and tools surge. 46:57 – Tesla FSD v14 • If safety drivers can be removed, Tesla’s model shifts from carmaker to AI mobility platform. • Success in vision-based autonomy directly accelerates humanoid deployment in factories and homes. 52:13 – AI & Bitcoin • Nvidia and Bitcoin cycles remain tightly correlated, reflecting capital’s linkage between compute and scarcity assets. • Labor market disruption will force policy responses (rate cuts, fiscal support), providing further tailwinds for BTC. 👉 Watch here