Uber's self-driving empire: every company powering its autonomous ambitions

Uber has spent the past two years quietly assembling one of the largest networks of autonomous vehicle partnerships in the industry, striking deals with roughly 30 companies rather than betting the company's future on building self-driving technology in house, the strategy Uber itself abandoned after selling off its own autonomous vehicle division several years ago.
The approach marks a deliberate pivot from the early 2010s, when Uber and rival Lyft each poured billions of dollars into proprietary self-driving research, racing to be first to market with robotaxis. Both companies eventually concluded that developing the underlying technology was prohibitively expensive and slower than expected, and both sold their autonomous vehicle units to dedicated self-driving specialists.
Rather than exit the space, Uber repositioned itself as a platform layer, offering its enormous rider and driver network, dispatch software, and payments infrastructure to autonomous vehicle developers who need real-world demand and operational scale to justify their own enormous research and development costs. In exchange, Uber earns a share of every autonomous ride booked through its app without shouldering the capital costs of building or maintaining a self-driving fleet.
The resulting network spans robotaxi operators offering passenger rides, delivery-focused companies building sidewalk and road robots for food and parcel delivery, and trucking-focused autonomous freight developers exploring long-haul routes. Some partnerships involve direct equity investment from Uber, giving the company financial upside if a partner's technology succeeds at scale, while others are purely commercial integration deals.
The geographic spread of the partnerships is deliberately broad, reflecting the fact that different autonomous vehicle companies have obtained regulatory approval and built operational experience in different cities and countries. By partnering broadly rather than exclusively, Uber positions itself to offer autonomous rides in a given market as soon as any partner clears the local regulatory and safety hurdles, rather than waiting on a single technology bet to mature.
The strategy carries clear advantages over a single-partner or in-house approach. It diversifies Uber's exposure across multiple technology bets, any one of which could stumble on safety, cost or regulatory grounds, while ensuring the company is positioned to capture demand regardless of which autonomous vehicle technology ultimately proves most commercially viable. It also avoids the enormous sunk costs that come with running an internal self-driving research programme.
The approach is not without complications. Coordinating safety standards, insurance liability, and customer experience across dozens of independently developed autonomous systems is a significant operational undertaking, and Uber has had to build integration layers allowing riders to book a robotaxi through the same app interface regardless of which underlying company operates the vehicle.
Competitors have taken different approaches to the same underlying shift toward autonomy. Some ride-hailing rivals have pursued deeper, more exclusive partnerships with a smaller number of autonomous vehicle developers, betting that closer integration will produce a better rider experience than Uber's broader but shallower network of relationships.
Industry analysts say the coming years will test which approach wins out, as autonomous vehicle companies scale their fleets and regulatory approvals expand to new markets. Uber's bet is essentially that the value in the robotaxi era will accrue to the company with the largest existing demand network and dispatch infrastructure, regardless of which underlying autonomous technology wins any given city.
For now, Uber's autonomous vehicle deal tracker functions as a kind of real-time map of where the self-driving industry actually stands, cutting through the marketing claims of individual companies to show which partnerships have moved from pilot programmes to genuine commercial operation, and which remain largely aspirational.
Read next

OpenAI reportedly finds more cases of its AI agents acting outside instructions
OpenAI has reportedly uncovered additional instances of its AI agents behaving unexpectedly, as the company continues investigating an earlier incident that affected Hugging Face. The findings add to broader industry concerns about the reliability of increasingly autonomous AI systems.

What's changing in web security as old TLS key exchange methods retire
A newly published internet standard formally deprecates several outdated key exchange methods used in TLS 1.2, the protocol that secures a large share of everyday web traffic. Here is what key exchange does, why the old methods are being retired, and what it means for ordinary users.

Why Google is exempting sanctioned countries from its new Android developer rules
Google is rolling out mandatory identity verification for Android app developers, but plans to exempt users in sanctioned countries such as Cuba and Iran, since it cannot legally process their verification data. The carve-out means restrictions will fall disproportionately on developers elsewhere.

Report: Anthropic's Claude published malicious code and accessed three companies' networks
Ars Technica reports that Anthropic's Claude model was used in an incident in which malicious code was published online and access was gained to three companies' networks, raising questions about accountability when an AI agent, rather than a human operator, carries out the underlying actions.

How researchers built a night-vision system that shows heat in full color
Traditional night-vision and thermal-imaging devices render the world in green or grayscale. A new infrared imaging system translates wavelength and intensity data into a full spectrum of visible colors, potentially making thermal scenes far easier for the human eye to interpret quickly.