Bernstein SocGen Group has maintained its Outperform rating and $95 price target on Uber, with the core rationale being that the company is accelerating the commercial deployment of autonomous vehicles.

Bernstein believes that "physical AI" is becoming a major new theme in the tech industry, and autonomous driving is one of the earliest applications to reach the commercialization stage.

Uber expects to launch autonomous vehicle services in 15 cities by the end of 2026, with 6 located in the United States and 9 in overseas markets.

Autonomous driving deployment is shifting from pilot programs to scale expansion

Uber is currently partnering with multiple autonomous driving technology companies rather than betting on a single supplier. Bernstein notes that key projects currently underway and upcoming include: Nuro's deployments in the San Francisco Bay Area and Houston, Zoox's projects in Las Vegas and Los Angeles, and Wayve's expansion in London and Tokyo.

Avride is also testing driverless services in Texas, though Bernstein points out that its safety record still needs further improvement.

From a partnership structure perspective, Uber is positioning itself as an aggregation and distribution platform for autonomous driving capacity, rather than taking on all autonomous driving technology development itself. This means that if different autonomous driving companies need to find passenger traffic and order entry points in the future, Uber has the opportunity to leverage its existing platform scale to become the core channel connecting autonomous fleets with consumers.

Nvidia's role in the autonomous driving ecosystem may continue to expand

Bernstein expects Nvidia to continue playing an important role in the autonomous driving ecosystem in 2027 and may further deepen its partnership with Uber. This reflects that Uber's autonomous driving strategy is not simply about connecting to Robotaxi operators, but gradually forming a broader technology collaboration network with chips, autonomous driving systems, and vehicle platforms.

If Uber can simultaneously connect to multiple autonomous driving technology stacks, its platform's dependence on any single supplier will be further reduced.

What the market is really worried about is whether Uber will be "bypassed" by autonomous driving

Bernstein believes that Uber's current valuation remains low, but there is a long-standing core concern in the market: once autonomous driving becomes fully widespread, will Robotaxi operators like Waymo and Zoox bypass Uber and operate directly facing consumers, thereby weakening Uber's long-term platform value.

This is also the so-called "terminal value bear case." If autonomous driving companies eventually control their own vehicles, technology, and user entry points, then Uber's traditional driver-passenger matching model could indeed face challenges.

Therefore, Bernstein believes that simply announcing more autonomous driving partnerships is not enough. What the market really needs to see is a relatively new autonomous driving company achieving genuine driverless commercial deployment in the United States through the Uber platform.

Once this model is proven, it would demonstrate that autonomous driving companies' technological capabilities and Uber's platform value do not necessarily replace each other.

Driverless deployment in the United States will be the most critical validation

Bernstein specifically points out that achieving commercial operation without safety drivers in the U.S. market by a new autonomous driving participant will be one of the most important data points going forward.

The reason is that this can answer a key question: after autonomous driving becomes widespread in the future, will Uber be weakened, or can it become the demand entry point commonly used by various driverless fleets?

If more and more autonomous driving companies choose to obtain orders through Uber, Uber's business model could evolve from "connecting drivers and passengers" to further becoming "connecting various types of capacity and passengers." Under this model, even if drivers gradually exit, Uber's network effects, user scale, and order distribution capabilities may still be preserved.

Bernstein believes the valuation has not yet reflected this potential

Bernstein believes that Uber's current valuation multiple is still "too cheap." But the firm also emphasizes that to truly eliminate long-term investors' concerns about autonomous driving disrupting Uber's business model, the company still needs to provide more actual driverless deployment data.

Therefore, the core of Bernstein's decision to maintain its positive rating this time is not that it believes autonomous driving poses no threat to Uber, but rather that Uber is transforming potential disruptors into new capacity on its platform through partnerships with multiple autonomous driving companies.

The most critical validation ahead will be whether Uber can achieve genuine driverless operations in the United States provided by a new-generation autonomous driving partner. Once this model is established, the market's discussion of Uber's long-term value may also shift from "whether autonomous driving will replace Uber" to "whether Uber can become the largest capacity aggregation platform in the autonomous driving era."