Leshka.eth ⛩ (@leshka_eth) “🚨THE AI TRADE IS OVER. CAPITAL IS FLOWING INTO ROBOTICS The last 3 years were a” — TopicDigg

🚨THE AI TRADE IS OVER. CAPITAL IS FLOWING INTO ROBOTICS The last 3 years were about LLMs, models, data centers, chips $700B in hyperscaler capex in 2026 alone The infrastructure layer got built Goldman flagged it directly: Investors are rotating away from AI infrastructure companies where earnings growth is under pressure and capex is debt-funded This narrative is already priced Nvidia is up only 12% in 2026 despite growing Data Center revenue 75% year over year The SMH semiconductor ETF, meanwhile, is up 74% The stock that should be winning the digital AI trade is getting faded Which means capital is ROTATING into the next layer To keep the rally alive, the market needs a new thesis That thesis is Physical AI Robots and autonomous systems that see, move, and operate in the real world Jensen Huang called it a $40 TRILLION market Today at Computex he announced Isaac Root Nvidia's first fully integrated humanoid robot running a Blackwell GPU and their entire software stack The biggest hedge funds already started accumulating positions: > BlackRock's official view: "humanoid robotics may be the most transformative frontier in physical AI, with the potential to reshape global labor markets and unlock trillions in value" They put the CEO of Agility Robotics on their board of directors this year > SoftBank is spinning out a $100B robotics IPO called Roze that is using robots to build AI data centers > VC funding into robotics hit $22B in 2025, up 69% year over year. Expected to double again in 2026 > Berkshire Hathaway tripled its $GOOGL stake in Q1 2026. The angle: Waymo - the most advanced self-driving platform in the world, operating commercial robotaxi fleets today Abel's first big tech bet landed squarely in Physical AI >Leopold Aschenbrenner dumped $8.46B in puts against $NVDA, $AMD, $TSM, and $SMH before And loaded in physical infrastructure: $IREN, $CORZ, $CLSK, $APLD, $BE, $VST His thesis: the chip layer is overpriced, the physical layer is where the real bottleneck is The next step of that same logic is robots or its infra The trillions poured into digital AI now fund the deployment layer The flywheel is starting