This Week in the Autonomy Economy, we mixed things up a bit as we turned our attention to London with the announcement that Baidu has begun testing their RT6 robotaxis on both the Freenow by Lyft and Uber networks.

We wondered what this meant for the market, so we asked our proprietary intelligence model, OMEGA, what it thought about the London robotaxi market. What came back impressed us, so we decided to dedicate this week's newsletter to that analysis.

For those of you who are interested, we will slowly begin opening OMEGA access this fall. If you are interested in learning more about how you will be able to access OMEGA, just reply to this email and we will be in touch.

Without further ado, here is Friends Today, Competitors Tomorrow.

WHAT’S MOVING THE MARKETS | AUTONOMOUS VEHICLES

Friends Today, Competitors Tomorrow

While the market debates the impact that the Uber / Waymo divorce will have on Uber's long-term robotaxi plans, the seeds for the divorce appear to have emerged very early in the marriage, as multiple media outlets are now reporting.

While media outlets report on the past, we are focused on the future. Our proprietary intelligence model, OMEGA, noticed on the London Assembly's Autonomous Passenger Vehicles in London investigation page that Waymo has a fleet of 100 vehicles in London.

100 vehicles in a market, testing across 20 boroughs from Barking & Dagenham to Wandsworth, is not a test or a deployment. It's a beachhead that signals grander ambitions.

In London, Waymo's geographic footprint, spanning the City of London and Westminster at the core out to Barking & Dagenham in the east, Ealing in the west, Haringey and Waltham Forest in the north, and Greenwich and Lewisham south of the river, gives us a sense of just how grand Waymo's London ambitions truly are. For Waymo, London is Calling.

It's not just Waymo London is calling. It's also Baidu, who has partnerships with both Freenow by Lyft and Uber. Baidu is currently testing an undisclosed number of vehicles in the borough of Brent, home to Wembley Stadium and roughly 45 minutes from Buckingham Palace, not exactly central London where Waymo is testing.

The permit framework under the Automated Vehicles Act 2024 came into force in May, with Automated Passenger Service (APS) permits issued nationally by the Department for Transport and Transport for London (TfL) in a consenting role.

There has been no official notice yet on when either Waymo or Baidu can launch commercial service and in what boroughs, but there is an interesting caveat that very few have pointed out, and it further gives Waymo a competitive advantage in the London market.

A path to an APS permit without a partner. There is no partner risk for Waymo this time. Perhaps Waymo learned that lesson in Austin and Atlanta? In London, Waymo is going to operate the service, own the asset, the app, and most importantly the APS permit.

As Gene Simmons told us years ago, it's your name on the door, so own it with pride. Indeed Waymo is owning it with pride, but most importantly they are removing a key risk that Wayve will be exposed to with their London Uber partnership and Baidu could be exposed to with both their Uber and Freenow by Lyft partnerships, unless Baidu flips the script and pulls a Waymo by holding the APS permit.

Wayve and Baidu did not partner with Uber and Freenow by Lyft for the demand generation narrative that has since been retired. They partnered for the license and the speed it unlocks. Uber and Freenow by Lyft both hold TfL private hire operator licenses, the regulatory asset that allows a passenger to be lawfully carried for hire in London today.

Testing autonomous vehicles does not require that asset, as all three programs run under the Department for Transport Trialling Code with safety drivers.

The license matters at the moment testing ends and revenue generation begins. Uber and Freenow by Lyft are licensed operators with years of standing at TfL, the consenting authority for APS permits covering taxi-like services in the capital, and the DfT fast-tracked its first passenger pilots through established operators.

For an autonomous vehicle company looking to enter the London market, the partner route is the shortcut, until the partner is no longer needed and the autonomous vehicle company secures their own APS permit.

Once an autonomous vehicle company holds their own APS permit and can operate a commercial autonomous vehicle service (robotaxi) under their own authority, the partner's value evaporates and they start becoming a toll, even as they continue to pitch the demand narrative to the market.

Demand is a commercial question, not a structural dependency. Structural dependencies are what hold partnerships together when interests diverge, and when that dependency is no longer needed, partnerships tend to dissolve and in some cases end in divorce.

Which raises the question, what does the demand partner do once their autonomous partner secures an APS permit in London? Search for a new partner, deploy their own owned-and-operated service or deploy a regulatory capture strategy to try and slow down the market?

It is an important question, as the APS permit is a durable regulatory asset. Once an autonomous vehicle operator secures one, they hold a national grant of commercial authority for half a decade, with a renewal window that opens six months before expiry and a provision keeping the permit valid while renewal is determined. Secure the permit, bye bye partner, as the value of a TfL private hire operator license evaporates the moment an APS permit is secured.

We have seen this movie play out before, tangentially, with Waymo and Uber, and now this scenario playing out again is probable with both Uber and Freenow by Lyft, as they are merely acting as the "London Bridge" for Baidu.

If Baidu applies for and secures their own independent APS permit, which we believe they will, the question becomes; why would Baidu want to continue to pay a toll when the partner's regulatory value is no longer there?

It's real and we said the quiet part out loud. The partner value is not there, as Baidu has the financial resources to launch a massive marketing campaign and deploy the Apollo Go app in London, followed by Europe, all without a partner, just as they have done in China. If they indeed went this route, we believe they would have no problem generating rider demand. After all, apps are replaceable and there is no true mobility moat yet.

To understand the London robotaxi market, you have to understand how the Automated Vehicles Act 2024 splits responsibility. The Act does not create one license. It creates three distinct legal designations, and who holds which one determines who controls the market.

Map all three entrants against the framework and their strategies stop looking similar and start looking like three different answers to the same question.

Waymo Holds All the Cards ASDE, NUiC Operator (with Moove as a contractor) and APS Permit Holder. Waymo's London structure has zero divorce risk, because there is no partner and no marriage.

This is the benchmark the other two alliances are measured against, and it is the structure the Act was arguably written to accommodate: a single accountable operator answering to the Secretary of State.

Wayve and Uber Split the Stack by Design Wayve is the ASDE and nothing more, while Uber will own the vehicles, operate the fleet, and most likely hold the APS permit. In London, Uber did not lend its regulatory standing to someone else's robotaxi service and hope the marriage would last decades.

Instead, Uber is poised to claim two of the three statutory designations for itself and leave its partner holding the one that cannot be monetized without a fleet. Wayve accepts that trade because licensing is its entire business model. The structure is stable for the same reason it is unequal. Each side of the deal holds something the other has chosen not to build. At least for now.

Baidu, Freenow by Lyft and Uber Left the Key Decision Unanswered The Act permits the APS Permit Holder to be the software developer, the fleet operator, or a third-party mobility provider. That flexibility is not a technicality. It is the fork in the road for both Freenow by Lyft and Uber's partnerships with Baidu, and it creates two possible paths forward. A company with two demand partners needs neither of them forever.

Option A: Baidu Holds the APS Permit Directly We believe this is the most likely path forward for Baidu. Under this scenario, Baidu acts as both the ASDE and the APS Permit Holder, stacking the safety authorization and the passenger service authority in one company.

Both Freenow by Lyft and Uber's roles would be reduced to demand aggregators until their initial agreements expire and Baidu launches the Apollo Go app in London. The strategic consequence of this scenario would be Uber's second high-profile robotaxi divorce and Lyft's first.

Once Baidu holds a national five-year permit directly from the Department for Transport, and the Apollo Go brand is recognizable in London, both Freenow by Lyft and Uber become a replaceable software layer.

Option B: Freenow by Lyft and Uber Both Apply as Permit Holders This is the least likely path in our opinion. Freenow by Lyft and Uber apply for APS permits as the designated passenger service providers, running the service on Baidu's RT6 platform under Baidu's ASDE safety clearance.

Both Freenow by Lyft and Uber would manage the passenger safeguarding, accessibility, customer support, and local fleet logistics, and in Uber's case with partners.

Under this scenario, both Freenow by Lyft and Uber would be valuable to Baidu. They would collectively manage the customer relationship and the national authority to serve them. But the moat has a drawbridge Baidu controls. Baidu owns the physical RT6 hardware and the autonomous driving stack.

If commercial terms turn unfavorable, Baidu can pull its vehicle allocation, and Freenow by Lyft would be left holding a permit to operate a fleet it no longer has, while Uber could add more vehicles to its Wayve-powered owned-and-operated robotaxi fleet.

The takeaway cuts one direction in each scenario. If Baidu holds the permit, Freenow by Lyft and Uber are replaceable apps. If Freenow by Lyft and Uber hold the permits, they keep the regulatory moat but stay dependent on Baidu for the hardware and the safety certification.

Either way, one party ends up holding leverage the other cannot replicate. Partnerships survive when leverage is balanced. These partnerships are not built that way.

Line the three structures up and the market reads clean. Waymo eliminated the counterparty. Uber captured the counterparty at the term sheet with Wayve.

Baidu, Freenow by Lyft and Uber, to some degree, deferred the question, and deferred questions in partnerships have a way of answering themselves at the worst possible moment.

While we wait to see how the Freenow by Lyft and Uber partnerships with Baidu evolve over time, Uber has already hedged their bet. Perhaps they learned this hedge during their tumultuous short-lived marriage to Waymo. Ownership.

That is the key difference in the Wayve partnership. Uber is going to own the asset and operate the fleet. Under the terms disclosed alongside Wayve's Series D, which Uber participated in with additional capital committed to multi-year deployments targeting more than 10 markets globally, with London being the first, Wayve deploys its AI Driver in L4-capable vehicles from participating automakers while Uber owns and operates the vehicles on the Uber network.

Uber is not a demand channel here. They listened to what the market told them. They are not a demand generation platform, they are now the asset owner and fleet operator. Wayve is the software licensor.

That is a fundamentally different arrangement than lending a TfL license to someone else's robotaxi, and it answers a question Uber has been living with since the Waymo relationship began deteriorating: what does Uber do when their robotaxi partner comes to the conclusion that Uber is no longer needed? The toll, aka the Uber tax, simply does not make business sense.

The answer, it appears, is build the capability to own and operate the fleets with a partner whose business model is licensing, not competing. Wayve does not want to own fleets. Wayve wants its AI Driver in as many vehicles as possible, whoever operates them.

Uber does not want to build an autonomous driving stack (at least for now). Uber wants fleets on its network that no partner can walk away with, lock them in and compete against Waymo.

Is Uber's Waymo hedging strategy coming to light, starting with London? Waymo and Uber were once friends, now they are competitors. Will the same happen with Baidu? If it does, does Uber further run into the arms of Wayve and Nuro? Then where does Lyft turn?

In London, Waymo holds all the cards, while Uber and Freenow by Lyft are about to learn just how valuable their platforms truly are, or are not.

Our Take: All eyes are on London and we will be heading to London at some point in the near future for a Field Report.

THE ROAD TO AUTONOMY ROBOTAXI INDEX

The Road to Autonomy Robotaxi Index (Top 5) | 18:00 UTC · 2026-08-02

📈 Follow @AutonomyIndices on X for the latest updates on The Road to Autonomy Indices. Updates every 12 hours.

AUTONOMY SIGNALS PODCAST

Tesla FSD Supervised Goes Coast to Coast as Baidu Robotaxis Arrive in London

(July 31, 2026) This week on Autonomy Signals presented by KPMG, Grayson Brulte and Rob Grant discuss the latest Tesla FSD Supervised Cannonball Run record, seniors adopting FSD as a mobility preservation tool, and Baidu beginning robotaxi testing in London on both the Uber and Lyft networks.

Watch on YouTube | Spotify | X

AUTONOMY MARKETS PODCAST

Is ADAS the Trojan Horse Into Autonomy?

(August 1, 2026) This week on Autonomy Markets, Grayson Brulte and Walter Piecyk discuss the Waymo/Uber divorce going public, Zoox receiving NHTSA approval to charge for rides, Aurora earnings, and Qualcomm's BMW win as ADAS becomes the Trojan horse into autonomy.

Watch on YouTube | Spotify | X