GM. This is Milk Road AI, built for the people who want to understand what's happening in tech before their group chat does.

​Here’s what we’ve got for you today:

Nexo is back in the U.S. - and new clients get 30 days of Wealth Club Premier perks! Higher yields, lower borrowing rates, and crypto cashback - start here.

THE ROBOTAXI WAR HAS A SILENT WINNER

During the most sensitive negotiations of World War II, the most powerful man in the room wasn’t the general or the diplomat.

It was the translator standing between them.

The guy who spoke both languages controlled what information moved, how it was framed, what the other side actually heard, and crucially, what they didn't.

Generals had armies, ambassadors had credentials, but the translator had leverage.

He was the only person everyone depended on, and the moment he was removed, the entire meeting stopped working.

Uber just spent 15 years becoming the translator between humans who need rides and the companies building the cars that will eventually drive them.

And now almost every robot car company on earth has to get through them to reach the passenger.

Most people are watching the wrong people in this room, and the data proves it.

Uber’s own data shows AVs (autonomous vehicles) on its platform complete ~30% more trips per day than standalone fleets.

And in this business, utilization is everything.

An AV fleet sitting idle between rides isn't resting, it's depreciating.

Every minute, a $150,000 robot car is parked waiting for its next fare, it's bleeding money like a faucet someone forgot to turn off. 

The vehicle still has insurance and maintenance costs, which is why Uber’s demand density keeps it earning instead of rusting.

And the math gets uncomfortable fast.

If a standalone robotaxi completes 4 trips per hour on its own, but 5.2 trips per hour on Uber's platform, then paying Uber a 20% cut still makes the AV operator more money than going solo. 

You're giving up a slice of every pie, but you're getting dramatically more pies.

This is exactly why Waymo, arguably one of the most sophisticated autonomous vehicle operations on the planet, chose to partner with Uber in Austin and Atlanta rather than go direct-to-consumer in those markets.

The smartest, best-funded, most technically capable AV company in the world looked at the data, ran the numbers, and decided they still needed the translator.

What Uber actually built 

Everyone thinks Uber’s moat is the little black car icon on your phone screen, but that couldn’t be further from the truth.

Here’s what Uber actually built: 

Reading that list is about as exciting as watching someone file their taxes.

Building it from scratch would cost a new entrant a decade of work and billions of dollars at a minimum.

Even then, they’d be starting without Uber’s behavioral data, regulatory relationships, or the habit of 202M people opening the same app every time they need a ride.

No AV startup can buy these capabilities off the shelf. 

Tesla isn’t licensing them, Waymo isn’t sharing, so every new AV company either builds all of it themselves or plugs into Uber to reach real, paying passengers, which is exactly why Uber formalized the whole thing into a product.

They launched a new division called Uber Autonomous Solutions, which packages all of this infrastructure and sells access directly to AV partners. 

Training data from Uber’s test fleet, mapping refined over billions of trips, real-time telemetry, demand generation, in-car software, and customer support infrastructure.

The full translator kit is available for licensing by any AV company that needs to reach riders without spending a decade building the relationships to find them.

Jeff Bezos built warehouses to sell books, then turned that infrastructure into AWS by renting compute to the world, creating the most profitable division in Amazon’s history.

Dara Khosrowshahi, CEO of Uber, built a demand network to move humans, and now he's renting that network to the robots that will eventually replace them.

THE ROBOTAXI WAR HAS A SILENT WINNER? (P2)

There is one man who has made it loudly and repeatedly clear that he is not paying anyone to translate for him.

His name is Elon Musk, and he believes in vertical integration the way some people believe in religion with the same absolute certainty and the same deep suspicion of anyone who suggests an alternative approach.

Tesla is not joining Uber's platform, and this is the part of this story that should genuinely keep Uber shareholders up at night.

Tesla isn't just a competitor building a slightly better AV stack and plugging it into the existing ecosystem. 

They are constructing an entirely parallel universe, hardware, software, chips, energy, insurance, and a consumer app that doesn't need a translator, a platform, or anyone's permission to operate.

Tesla makes its own cars, runs its own chips, and trains its AI on billions of miles of real-world driving data from every vehicle it has ever sold, something no competitor can replicate from a standing start.

That vertical integration gives Tesla unmatched control over both cost and performance, and the industry data is starting to reflect that shift.

According to ARK Invest estimates, robotaxi pricing could fall toward ~$0.25 per mile at scale.

Even Waymo’s cost curve points in the same direction as it scales, just not as aggressively as Tesla’s fully integrated model.

This is where the gap really shows up.

There is no LiDAR hardware cost, no third-party autonomy licensing fees, and no non-captive insurance expenses, every dollar Uber’s model has to share across partners, Tesla keeps.

That structural advantage is why a Tesla robotaxi in San Francisco was recently averaging $8.17 per ride while Lyft averaged $15.47 for the same trip across town.

Half the price, same destination, and no translator required.

The portfolio play nobody is writing about

Here is what makes Uber's position genuinely interesting despite the Tesla problem, and why the story is more complicated than Elon wins and everyone goes home.

Uber is spreading its bets across the entire race, not backing a single winner.

While every financial media outlet is laser-focused on the Waymo vs. Tesla main event, which is admittedly a great story with two compelling characters.

Meanwhile, Uber has quietly built the most comprehensive autonomous vehicle partnership network in the world, and barely anyone is talking about it that way.

Waymo is live and commercially operating on Uber's platform in Austin and Atlanta.

Rivian has committed to up to 50,000 vehicles across 25 cities through 2031, with the first 10,000 already locked in.

NVIDIA’s autonomous platform is rolling out across 28 cities starting next year, while Nuro and Lucid are targeting San Francisco in 2026.

Volkswagen’s ID. Buzz is in testing in Los Angeles, WeRide is already live with commercial trips in Abu Dhabi, Dubai, and Riyadh, and May Mobility is operating in Arlington, Texas today.

Twenty partners across four continents, spanning every major approach from camera-only and LiDAR-heavy systems to neural networks and traditional SLAM (simultaneous localization and mapping).

Uber’s recent $1.25B deal with Rivian tells you everything about how it’s approaching this.

They didn't just write a check and wish Rivian good luck. 

They structured milestone-based capital tranches, meaning later rounds of investment only deploy if Rivian actually hits specific autonomous performance benchmarks, protecting Uber's capital if the technology underdelivers. 

They locked the entire Rivian fleet to Uber's platform exclusively across those 25 cities. 

They secured the option on 40,000 additional vehicles starting in 2030.

And they structured the deal so that Rivian gets a committed commercial path without having to build its own consumer app, payment processing, routing algorithms, or demand generation from scratch.

And the strategic logic is genuinely elegant in a way that doesn't get enough credit: if Waymo's technology proves superior, Uber benefits. 

If Nvidia's platform enables a whole new wave of AV entrants who couldn't previously afford to build full-stack systems, Uber benefits. 

If Nuro cracks the premium market segment, Uber benefits.

If a startup nobody has heard of yet figures out a breakthrough approach in 2027, Uber benefits.

The only scenario where Uber truly loses is one where AV supply consolidates entirely around a single vertically integrated player with the capital, brand, and consumer app to generate its own demand without any help.

The honest verdict

So what is Uber actually becoming? Something much closer to the next Visa.

Visa is not where JPMorgan wants your transaction fees going. 

The largest, most powerful banks in the world have been trying to route around Visa's rails for decades, building their own payment networks, lobbying for regulatory intervention, and occasionally just quietly seething about the whole arrangement over expensive dinners.

But the rest of the financial system still runs on Visa's infrastructure. 

Why? Building a credible alternative costs more than the platform charges to use, and the network effects mean that even a technically superior alternative starts at zero while Visa starts everywhere.

That is the exact position Uber is trying to cement before the AV era fully arrives and the leverage shifts.

And they are not waiting around for it to happen to them.

Uber Autonomous Solutions is the moment they stopped being a ride-hailing company and started being infrastructure. 

They are now packaging 15 years of demand data, regulatory relationships, and operational know-how and licensing it to the very companies that were supposed to disrupt them.

Real-time fleet telemetry, charging infrastructure, in-car software, and dynamic routing tuned across 50 billion real-world trips.

The full suite of tools needed to actually run a fleet of autonomous vehicles at scale not in a test environment, but in real cities, with real passengers who leave bad reviews when something goes wrong.

Most AV companies can build a car that drives itself. 

Almost none of them have figured out how to run a fleet of ten thousand of them across fifty cities simultaneously without the whole thing turning into an expensive logistics nightmare.

Uber has been solving that problem since 2009.

Tesla might route around it, but the other 18 AV companies with robot cars, brilliant engineers, and absolutely no consumer base, no pricing infrastructure, and no regulatory playbook across 70 jurisdictions?

They need the distribution layer, the one entity that speaks both languages, and right now, there is really only one translator available, and that’s Uber.

Alright, that’s it for this edition of Milk Road AI. We want to hear from you.

BITE-SIZED COOKIES FOR THE ROAD 🍪

OpenAI raised $122B at an $852B valuation, including $3B from retail investors. The massive round signals momentum as it gears up for a likely IPO.

Meta lost two major cases, finding it liable for harming teens through addictive app design. The rulings could trigger a wave of similar lawsuits over teen safety.

Google is rolling out a feature letting U.S. users change their Gmail addresses. Old emails stay intact, and the change is limited to once per year.

MILKY MEMES 🤣

ROADIE REVIEW OF THE DAY 🥛