Oppenheimer Maintains Neutral Rating on Tesla, Citing Shift to Fully Integrated Physical AI Platform
Oppenheimer reiterated its Perform rating on Tesla, citing the company's shift to a fully integrated Physical AI platform, moving away from pricing based solely on the vehicle story.
The brokerage noted that Tesla's execution in extending vertical integration and improving key supply chain aspects is acceptable, and described its self-developed semiconductor manufacturing as a bold yet crucial step: optimizing performance for high-leverage components while reducing exposure to supply chain risks from China's geopolitical expansion. The institutional knowledge accumulated by Dojo is said to determine whether Tesla can iterate chips ahead of its peers in various forms.
In line with this logic, capital expenditure is viewed as a leading indicator of whether Physical AI will succeed. The firm has previously raised its capital expenditure expectations for 2026 and 2027 to $29.4 billion and $30.4 billion, respectively, primarily due to upward revisions in data center and semiconductor spending, with the 2026 estimate being about 25% higher than the market. Whether power capacity can be launched on schedule and whether the learning cycle can accelerate are considered more critical observation points than quarterly vehicle sales.
On the product side, innovations in Cybercab manufacturing and computing power expansion are highlighted, while it is noted that vertical integration will take years to close the loop, and the timelines for Cybercab and Optimus may continue to be pushed back. In 2026, free cash flow may turn negative again due to factory construction, and the Robotaxi fleet may operate off-balance sheet. Valuation still corresponds to an extremely high price-to-earnings ratio, with cash on hand exceeding debt.
Buyers are revaluing Tesla as a growth fund for an energy-chip-automobile-robotics closed loop; sellers are focused on the profitability of the vehicle business and the timing of cash realization. The event-driven aspect is that the rating remains unchanged but the narrative has been rewritten: a neutral rating paired with platform theory means neither chasing highs nor downgrading. Beneficiaries are those exposed to chips, energy storage, and autonomous driving software; those under pressure are shorts still valuing Tesla based on the automotive cycle.
The market is trading on how much usable computing power, electricity, and learning speed can be obtained for every dollar of capital expenditure, not just another maintained rating.
Source: Public Information
ABAB AI Insight
Oppenheimer has already issued a Perform rating on Tesla: earlier this year, it indicated that the shift to Physical AI would drag down Cybercab and Optimus, predicting negative cash flow in 2026, and mid-year changed the narrative to "watch if capital expenditure leads to success." Not giving a buy rating but only a neutral one acknowledges that the story has changed, but they are unwilling to sign off at a high price-to-earnings ratio. This aligns with the analyst's public statement after the quarterly report that "we haven't seen results from the shift to physical AI yet": the narrative comes first, and the financial statements follow.
The capital path involves extracting gross margins from vehicles to repair chips, data centers, and energy storage, using Dojo's experience to create various forms of chips, and treating the Cybercab production line as the first mass production container for physical AI. The motivation is to shift the highest leverage components from external procurement to self-development, hedging against Asian supply chain and political risks. Resource mobilization is reflected in continuous upward revisions of capital expenditure and possibly moving the Robotaxi fleet off-balance sheet to avoid putting factory investments and fleet depreciation on the financial statements simultaneously.
Comparative entities include NVIDIA selling computing power, Google making TPUs, Amazon producing Trainium, and traditional automakers purchasing chips while treating driver assistance as an optional package. Tesla aims to do cars, electricity, chips, and robots simultaneously, positioning itself as a "platform integrator" rather than a single model company. SpaceX is viewed by the same group of analysts as a neighboring entry point for physical AI, with mergers still seen as an option years down the line.
The structural judgment belongs to the overlay of technological substitution and capital concentration. For software intelligence to translate into steering wheels and limbs, electricity, chips, data, and factories must be consolidated into the same balance sheet. The mechanism is: the speed of the learning cycle depends on the data generated by real vehicles and robots, as well as the ability to produce matching chips in-house; whoever controls these three will no longer price based on vehicle count. Maintaining the Perform rating indicates that Wall Street accepts the narrative switch but still requires seeing capital expenditure turn into measurable computing power and delivery, rather than just slogans.
ABAB News · Cognitive Law
Capital expenditure will precede profits, becoming the new story's test.
Vertical integration buys risk hedging, but the cost is realization time.
A neutral rating paired with a new narrative equals acknowledgment of a shift while refusing to chase highs.