Everyone and their brother is playing the AI trade through the exact same lens: chips, cloud hyperscalers, and chatbots.
Don't get me wrong, NVDA and MSFT are selling the premium picks and shovels. But I'm becoming way more interested in the orchestration and application layer—specifically, what happens when AI starts moving into the physical world to fix actual operational bottlenecks.
I’ve built a small, highly speculative sleeve of my portfolio via LEAPs targeting this "AI meets atoms" thesis. These are volatile, mostly unprofitable, and richly valued. They could absolutely crater (very high probability but this is WSB) but if AI genuinely transforms the physical economy over the next decade, the asymmetric upside here is massive.
Here is the current basket I'm holding:
* AUR (Aurora Innovation) — Autonomous Trucking. If self-driving freight works at scale, this completely re-engineers global supply chain economics. The data loops they are building on actual highway miles create a massive moat.
* SYM (Symbotic) — Warehouse Robotics. AI-driven automation inside massive distribution networks. They aren't just selling software; they are re-architecting physical fulfillment for giants like Walmart.
* IOT (Samsara) — Industrial IoT & Fleet Intelligence. They are turning unsexy physical assets (trucks, heavy equipment, job sites) into real-time data platforms. The data verticalization here is incredibly sticky.
* RXRX (Recursion Pharmaceuticals) — AI Drug Discovery. Applying ML to biological research to shrink the time and failure rate of bringing a drug to market. High risk, but pure structural disruption.
* TEM (Tempus AI) — Precision Medicine. Building massive, proprietary datasets in healthcare for clinical decision support and diagnostics.
The core thesis is simple: The companies that win the next phase of AI won’t win because they have the prettiest LLM wrapper. They’ll win because they own the closed data feedback loops in specific, high-friction physical verticals.
I’m looking to expand this basket with a few more non-obvious plays that are quietly plugging machine learning into physical infrastructure, logistics, or deep-tech operations.
What under-the-radar "physical AI" plays are on your watchlist right now? (And please, spare me the NVDA/PLTR standard talking points).