Morgan Stanley’s Jonas: Physical AI Could Multiply Global GDP

Morgan Stanley analyst Adam Jonas said convergence between AI and robotics could transform the global economy and potentially multiply GDP as physical AI expands across transportation, manufacturing and daily life. He highlighted China’s supply-chain advantage in robotics and argued that rebuilding US manufacturing is integral to the competitive race. Jonas sees Tesla and SpaceX as potentially complementary participants in a future of billions of connected robots conducting AI inference at the edge.

Analysis

The investable bottleneck in physical AI is unlikely to be model capability; it is bill-of-materials localization, safety validation, and reliable field-service infrastructure. Tesla’s valuation can gain further optionality if autonomy-derived compute, vision data, battery systems, and actuator manufacturing become reusable across vehicles and humanoids, but this remains a multi-year execution thesis rather than a near-term earnings driver. The more immediate read-through is to industrial automation suppliers with real revenue exposure to sensors, motion control, machine vision, and factory integration—areas where deployment budgets can materialize before general-purpose robots achieve scale.

China’s manufacturing density creates a second-order risk to a simple “US reshoring equals US robotics winner” narrative. A tariff-led localization push could initially raise capex and component costs for US adopters, depressing returns on automation projects even as it supports domestic equipment demand; meanwhile, Chinese firms can use their home-market scale to compress global robot hardware pricing. This favors software, controls, and high-switching-cost automation vendors over undifferentiated hardware assemblers.

Near term, the commentary is unlikely to alter TSLA fundamentals absent disclosed Optimus production volumes, external customer commitments, or separate segment economics. Over 1-3 months, investor attention will rotate around AI/robotics presentations and any evidence that Tesla can convert internal pilots into measurable labor savings; over 6-18 months, the relevant falsifier is whether robot programs remain R&D expense without a credible manufacturing ramp. SpaceX-related exposure should be treated as narrative-sensitive because SPCX is not a standard publicly traded operating-company ticker and offers no direct liquid public-equity expression of SpaceX economics.

Consensus is likely over-indexing on humanoid-unit forecasts and underweighting the adoption curve: factories buy uptime, integration certainty, and payback periods, not embodied-AI demonstrations. The first durable winners may therefore be ABB, Rockwell Automation, FANUC, Keyence, Cognex, and NVIDIA rather than the company with the most compelling humanoid prototype. A broad robotics rerating requires evidence of orders and gross-margin durability, not demonstrations or analyst GDP scenarios.

Key Decisions for Investors

• Maintain TSLA as a tactical, catalyst-driven long only if management discloses external Optimus orders, production targets, or quantified internal labor savings within the next 1-3 quarters; size as optionality, not a core robotics earnings position. Exit/reduce on a material increase in robotics R&D without corresponding production milestones or if automotive gross-margin guidance deteriorates, overwhelming the robotics narrative.
• Express the nearer-term physical-AI capex theme through a basket long ABB, ROK, CGNX and NVDA over 6-18 months, favoring controls, machine vision and compute exposure over pure humanoid narratives. Target a 15-25% basket upside from capex/order acceleration; reassess if US manufacturing PMIs and automation order books fail to improve over two consecutive quarters.
• Pair long ROK or ABB against short a broad low-differentiation industrial-hardware proxy only after evidence of tariff-driven component inflation emerges. The thesis is that pricing power and installed-base software/service revenue protect leaders while commodity automation hardware absorbs margin pressure; do not initiate without supplier-cost and backlog data.
• Avoid using SPCX as a direct SpaceX position until the instrument’s legal entity, liquidity, and economic linkage are independently verified. Treat any price move tied solely to SpaceX/robotics commentary as a liquidity and narrative-risk alert rather than a fundamental signal.