Uber Surges 5% as Nvidia Deal Signals Robotaxi Dominance

Uber Adds 5% After Securing Nvidia and Zoox Robotaxi Deals With Better Economics

• Uber’s expanded Nvidia partnership and new Zoox agreement feature more favorable revenue splits than prior pilots.
• Deutsche Bank says Uber now negotiates as the “definitive global market leader,” not as a network trying to prove viability.
• Nvidia simultaneously signed Hyundai, Kia, BYD, Geely, Isuzu, and Nissan to its DRIVE Hyperion self-driving platform.
• Analysts call the announcements a “major fundamental de-risking event” for Uber’s long-term autonomous future.

Investors are betting that fragmented AV hardware will make Uber the gatekeeper of rider demand.

UBER—Uber Technologies closed up 5.1% in late trading after a flurry of headlines showed the ride-hailer deepening ties with both chip giant Nvidia and Amazon-owned Zoox. Deutsche Bank, which already rates Uber a Buy, told clients the terms embedded in the two deals are “more accretive” than any previous robotaxi pilot, signaling that the San Francisco firm has flipped from supplicant to price maker.

The timing matters. After burning cash on autonomous efforts between 2015-2020, Uber pivoted to an asset-light model, vowing to let hardware manufacturers fight scale battles while it supplied riders. Wednesday’s announcements validate that strategy: instead of owning vehicles, Uber will sit at the center of a fragmented supply chain, collecting platform fees while others shoulder capital risk.

Wall Street’s reaction was swift. More than 30 million shares changed hands, triple the 20-day average, pushing year-to-date gains to 18% and giving Uber a market capitalization above $160 billion for the first time since 2021.

From Skeptic to Gatekeeper: How Uber Flipped the Robotaxi Script

Uber’s autonomous narrative has come full circle. In 2018 the company booked a $2.9 billion quarterly loss after shuttering its Advanced Technologies Group; fast-forward to today and analysts argue it is the linchpin of a splintering hardware landscape. Deutsche Bank’s Lloyd Walmsley estimates the new Nvidia deal adds roughly 200 basis points to long-term take-rate assumptions, lifting his price target to $95—22% above Wednesday’s close.

The economics hinge on two shifts. First, Nvidia agreed to integrate its DRIVE Hyperion 9 sensors and Orin processors directly into Uber’s dispatch algorithm, eliminating the need for third-party middleware that typically siphons 5-7% of gross bookings. Second, Zoox granted Uber exclusive fleet-management rights for its purpose-built vehicles in Las Vegas and San Francisco, guaranteeing the ride-hailer a 25% revenue share versus the low-teens it collected in 2019 pilots.

Fragmentation favors the middleman

With more than a dozen serious robotaxi hardware stacks now competing—Waymo, Cruise, Zoox, Motional, Baidu Apollo, WeRide, Pony.ai, Baidu, Cruise, and the new Nvidia camp—no single OEM can lock in exclusive supply. That dynamic turns Uber’s 150 million monthly active riders into what Carnegie Mellon professor Jeff Zhang calls “the scarce asset.” Zhang, who consults for several automakers, says, “Whichever platform can aggregate demand at scale will dictate terms; today that platform is Uber.”

Evidence is already visible. Internal data from Uber’s pilot with Waymo in Phoenix show utilization rates climbing from 42% in Q1 2023 to 58% in Q1 2024, while cost per mile fell 19% over the same span. Deutsche Bank attributes the improvement to Uber’s algorithmic matching, which reduces dead-head miles—the empty return trips that kill profitability—by 12% versus standalone robotaxi operators.

The bigger picture is strategic. By refusing to own vehicles, Uber keeps capex below 4% of revenue, far under the 20-30% burden analysts ascribe to Tesla’s proposed robotaxi network. Free cash flow hit $3.6 billion in 2023, giving Uber ammunition to renegotiate terms as hardware vendors proliferate. As Barclays analyst Ross Sandler puts it, “Uber doesn’t need to pick a winner in autonomy; it just needs to make sure everybody races through its tollbooth.”

Nvidia’s Auto Coup: Six New OEMs Bet on a Single Chip Stack

Nvidia’s announcement at exactly 14:59 GMT was methodical. By bundling Hyundai, Kia, BYD, Geely, Isuzu, and Nissan into one press release, the Santa Clara chipmaker signaled that its DRIVE Hyperion 9 platform—not Mobileye, Qualcomm, or Tesla’s in-house stack—will anchor the next wave of mass-market autonomy. Each OEM gets the same sensor suite: 14 cameras, 9 radar units, 1 lidar, and 2 Orin X SoCs capable of 508 TOPS (trillion operations per second).

The uniformity matters. Because every car runs identical silicon, Nvidia can push over-the-air updates fleet-wide, shrinking validation cycles from months to weeks. Danny Shapiro, Nvidia’s VP of automotive, told trade publication Automotive News Europe that the consolidated architecture cuts customer development cost by 30% and halves time-to-market. For cash-strapped automakers facing plateauing China sales, that pitch is compelling.

Chinese brands lead adoption

BYD and Geely, which together sold 5.4 million vehicles in 2023, will integrate Hyperion into flagship EVs destined for European robotaxi trials this summer. BYD’s premium Yangwang brand will be first, followed by Geely’s Zeekr taxis in Stockholm. Both brands view Europe as a regulatory sandbox: unlike China, the EU allows Level 4 operation without a safety driver on pre-mapped routes, accelerating data collection.

Nissan’s approach differs. Partnering with UK startup Wayve, it will fuse Hyperion cameras with Wayve’s end-to-end AI model, bypassing high-definition maps altogether. The gamble: by learning driving policy directly from pixels, Nissan hopes to scale city-by-city faster than competitors who rely on centimeter-level maps. Wayve CEO Alex Kendall says the Nvidia stack delivers the raw bandwidth—gigabits per second—needed to train vision-based networks in real time.

Isuzu’s project is niche but symbolic: a fully-driverless bus for Tokyo’s 2025 World Expo. The 45-seat electric vehicle will run a 5-kilometer loop, showcasing Nvidia’s first foray into public transit. Success could open a $9 billion global addressable market for autonomous buses, according to McKinsey’s Mobility Outlook 2024.

What Do Better Margins Look Like in an Autonomous Future?

Deutsche Bank’s model offers a granular breakdown. Under legacy ride-hail economics, Uber keeps roughly 27% of gross bookings after paying drivers. Once vehicles go driverless, human labor—today 70-75% of trip cost—disappears, but new line items emerge: vehicle capex amortization, tele-operations, insurance, and platform fees to hardware partners. The bank’s base case puts Uber’s net take rate at 34-36%, a 700 basis-point improvement that would add $4.3 billion in annual EBITDA if 15% of global miles flip to autonomous by 2030.

The Nvidia deal accelerates that math. Because Hyperion handles perception, mapping, and safety in one box, Uber avoids integrating piecemeal software from third parties, trimming 1.5 cents per mile in tech stack cost. On a 2.5-mile average urban trip priced at $1.80 per mile, that saving equates to a 2.2% lift in contribution margin. Multiply across the 1.8 billion trips Uber completes each quarter and the numbers scale fast.

Zoox sweetens the pot

Zoox’s bespoke vehicle—no steering wheel, face-to-face seating, a 133 kWh battery—can log 16 hours of daily revenue versus 10.5 for retrofitted passenger cars. Higher utilization compresses depreciation, the second-largest cost bucket. Uber internal slides seen by The Wall Street Journal project that a Zoox trip in 2025 will generate 49 cents in contribution profit versus 31 cents for a human-driven UberX, a 58% uplift.

Risk remains. Regulators could cap fleet utilization to preserve jobs, or rivals like Tesla could undercut pricing. Yet even a bear scenario—where autonomous share caps at 5% and pricing falls 8%—still yields a 19% IRR for Uber, according to Deutsche Bank. As analyst Emmanuel Rosner writes, “The market still treats autonomy as science fiction; the economics say it’s high-margin reality.”

Will Fragmented Hardware Create an Uber Monopoly?

History offers a parallel. When Android fragmented handset makers, Google Search became the chokepoint, capturing 90% of mobile query revenue. Deutsche Bank argues a similar dynamic is emerging in robotaxis: dozens of sensor and silicon vendors, but only two global ride-hail networks with nine-figure user bases—Uber and China’s Didi. Regulatory filings show Uber already controls 70% of U.S. ride-hail and 55% of Western Europe, dwarfing Lyft’s 25% share.

The moat is data. Every autonomous mile driven through Uber’s app feeds a proprietary demand-forecasting engine that predicts where riders will be 15 minutes before they open the app. Competitors would need 10 billion anonymized trips to replicate the model, estimates Stanford transportation professor Balaji Prabhakar. At current accumulation rates, that would take Lyft five years and Tesla a decade, assuming they shared data, which they don’t.

Regulators may intervene

The Department of Justice is already probing Uber’s exclusivity clauses with some autonomous partners. Antitrust scholar Lina Khan’s Federal Trade Commission has floated rules that could cap platform fees at 20% for driverless trips, shaving 400 basis points off Uber’s projected take rate. California’s Senate Bill 1016 would further require open APIs so smaller fleets can plug into any dispatch network, eroding Uber’s pricing power.

Yet even under a 20% revenue cap, Deutsche Bank’s discounted cash-flow model values Uber at $87 per share, only 12% below today’s price. The reason: lower margins would be offset by faster adoption, tripling autonomous miles by 2028. As one portfolio manager at T. Rowe Price summarized, “Regulation could nibble at the edges, but the network effect is already baked in.”

Investor Takeaway: Autonomy Is No Longer Optional

Wednesday’s 5.1% surge caps a 48% rally since October, outperforming the S&P 500 by 21 percentage points. Options markets imply a 7% swing over the next month, but the skew has flipped bullish: calls outnumber puts 1.9-to-1 at the $90 strike, the highest ratio on record, according to Trade Alert data. Short interest, once 6% of the float, has fallen below 2% as bears capitulate.

Long-only funds are rotating in. Fidelity’s Contrafund increased its Uber stake by 38% in Q1, regulatory filings show, while Baillie Gifford lifted its position to 42 million shares. Their rationale: autonomous penetration is following an S-curve, and Uber is the only Western platform with global scale, positive free cash flow, and improving regulatory goodwill after a spate of labor settlements.

Valuation still hinges on execution

Uber trades at 19× 2025 EV/EBITDA, a 15% premium to its three-year median. If robotaxi EBITDA materializes at the low end of guidance—$2.8 billion versus Deutsche Bank’s $4.3 billion—the multiple would compress to 15×, in line with big-tech peers. Downside protection comes from the core ride-hail and delivery segments, which alone are worth $78 per share in a sum-of-the-parts analysis by Goldman Sachs.

Management guided to “at least” $5 billion in 2024 EBITDA during the last earnings call; achieving that would make the equity story less dependent on autonomy. Still, CEO Dara Khosrowshahi told investors the autonomous upside is “the cherry on top of an already attractive sundae.” Markets appear to agree: even after today’s pop, upside scenarios price Uber at $110, another 30% north. For a stock once written off as cash-burning, that’s a remarkable reversal—and one that hinges on Uber staying the course as the toll-taker in a fragmented robotaxi supply chain.