Meng Xing Reviews Tesla’s 2024 Robotaxi Event

In October 2024, Zhang Xiaojun and Meng Xing discussed Tesla’s Robotaxi event, including Cybercab, Robovan, and Optimus. Their conversation captures judgments about the demonstration and plans at the time, not subsequent deployment results.

Meng thought the event felt more like a gathering for supporters: the atmosphere was enthusiastic, but it offered fewer technical details about autonomous driving than he had expected.

He observed that supporters and skeptics often retained their previous views. The absence of a demonstration alone cannot establish whether a company has a capability.

They discuss the event format and experience. Meng says Tesla had held short events before, but the lack of detail on software progress was a source of disappointment for him.

He divides the announcements into three: the two-seat Cybercab, the larger Robovan, and the Optimus humanoid robot. Dates for production and driverless services were plans announced at the event, not guarantees.

Discussing FSD, Meng distinguishes driver-assistance systems requiring supervision from automated driving that does not require continuous driver attention under specified conditions. He says naming, capability, and responsibility should not be conflated.

Asked whether a single vehicle could offer different driving modes, Meng says it is possible in theory, but depends on actual system capability, hardware, operating conditions, and responsibility.

He considers the event’s proposed 2025 timeline for unsupervised FSD ambitious and says the operating area, conditions, and actual performance matter. This was his 2024 assessment.

On increasing vehicle utilization, he says some benefits come from shared mobility rather than autonomous driving alone; safe driverless operation could also change the cost structure.

He treats greater safety as a goal and possibility, emphasizing complex scenarios, generalization, and reliable evidence rather than a conclusion from one demonstration.

Meng says the presentation did not provide enough information to determine the details of a technical change. A low-cost vehicle still requires separate assessment of hardware, software, and service readiness.

He reviews the Cybercab’s two-seat layout and lack of conventional driving controls, while questioning whether accessibility, maintenance, and cleaning needs for a ride service were fully addressed.

They discuss rumors and personal speculation that the vehicle was derived from another lower-cost model; the episode provides no evidence confirming its development history.

Meng offers another interpretation: when real operational experience is limited, starting with an existing prototype and iterating from user feedback may be one option.

He distinguishes closed-course testing, fixed routes on public roads, and broader open environments. Each introduces more traffic and unexpected situations, so a closed demonstration is not equivalent to broad deployment.

They discuss the event’s proposed per-mile operating cost. Meng argues that removing driver costs does not eliminate depreciation, maintenance, insurance, dispatch, and remote-support costs.

He offers a rough estimate and notes that labor and operating costs differ between the US and China. The estimate depends on assumptions and is not an actual quote or investment basis.

On the idea of an owner managing multiple driverless cars, he compares it with fleet ownership and platform operation, considering how capital, risk, and returns might be allocated.

He describes vehicle partners in ride-hailing: some companies own fleets and provide vehicles to drivers. Financing costs and management capability in the existing system affect whether a new model can work.

Comparing China and the US, he says individual participation in fleets may take different forms depending on local regulation, financing, and demand.

Asked about less visible operational work, Meng lists parking, cleaning, repairs, charging, and passenger issues. Even if a vehicle drives itself, those tasks still need arrangements.

He discusses charging and battery swapping: automated facilities, centralized depots, and third-party networks each have costs and constraints beyond the drive itself.

They discuss the share-price reaction after the event and suggest investor expectations about Tesla’s competition with ride-hailing platforms may have shifted. This is an interpretation of that moment’s market reaction.

Why pursue Robotaxi? Meng says that if the technology and service work, a company might earn recurring service revenue beyond vehicle sales. He acknowledges that he cannot know all of Tesla’s motives.

He sees potential in using parked vehicles for distributed computing, but networking, scheduling, and task suitability would need to be addressed. Simply adding advertised compute does not establish that a usable supercomputer exists.

On Robovan, he says a shared vehicle serving more passengers has different demand and economics from a taxi; operating costs, route density, and public subsidies affect viability.

He notes that companies have tested or operated autonomous shuttles in particular areas. Fixed-route demand can be easier to define, though market size and service area may be more limited.

On Optimus and autonomous driving, he says data infrastructure and some perception techniques may transfer, but household tasks and driving differ, so complete transfer of capability cannot be assumed.

He found the robot’s event interactions impressive but speculated that some movements might have involved remote human operation. The speakers did not independently verify the control method, so this remains their speculation.

On robotics research, he cites imitation learning and lab demonstrations, warning that smooth clips do not establish reliability when conditions change. Failure rates and generalization matter.

In closing, he distinguishes showing a vehicle from deploying driverless service at scale and remains skeptical of the announced timelines. The robot’s actual demonstrated capabilities also need further evidence.

He again characterizes the event as a presentation of future scenarios for existing supporters. The host treats it as a starting point for following product, technology, and commercialization progress.