Bernstein: Uber's New CFO Reveals AV Timeline Reality Check and Delivery Hero Stake Strategy
Bernstein 42nd Annual Strategic Decisions Conference, May 28, 2026 — Uber CFO Balaji Krishnamurthy's first major public appearance since appointment
Uber's newly minted CFO Balaji Krishnamurthy used his debut on the institutional conference circuit to deliver two messages investors have been waiting to hear clearly: the company has quietly built a 37% stake in Delivery Hero at a cost basis in the low thirties, and autonomous vehicles, despite all the enthusiasm, will remain largely immaterial to Uber's global volumes for at least five years. Neither point had been articulated with this degree of precision before.
Delivery Hero: A Seat at the Table, Not Yet a Deal
The Delivery Hero situation dominated the opening of the Bernstein conversation. Krishnamurthy confirmed that Uber now holds approximately 37% exposure in the company, built at a cost basis in the low thirties, and was careful to frame this as optionality rather than a committed acquisition. "With the position we now have, we have a seat at the table for the outcomes that may emerge from the strategic review of Delivery Hero," he said. Delivery Hero has publicly acknowledged it is reviewing strategic alternatives.
The strategic logic Krishnamurthy outlined centers on geographic complementarity rather than pure delivery scale. He pointed specifically to the Middle East, where Uber has a dominant Mobility presence but no Delivery footprint, as the clearest example of where a combination creates something genuinely additive. Markets like Argentina and Korea were also mentioned as organically strong positions that could fit. The CFO was equally clear about the financial discipline required: any transaction would need a credible path through regulatory approval, a concrete integration plan, and the ability to drive adjusted EPS accretion that justifies not simply buying back Uber's own stock — a comparison he made explicitly. "We don't want to overpay for acquisitions, even if it meets all the other criteria for us," he said.
What Krishnamurthy did not provide was any indication of where Uber's walk-away price sits relative to seller expectations, beyond noting that "for anything to happen, it would have to be at a seller expectation that's reasonable."
AVs: A Five-Year Reality Check That the Market Needs to Hear
The most consequential disclosure for longer-duration investors was Krishnamurthy's frank assessment of the AV contribution timeline. Asked directly about the five- and ten-year outlook, he stated that over a five-year horizon AVs will remain "relatively immaterial" relative to Uber's global and even U.S. volumes. The arithmetic he offered is instructive: Uber currently runs roughly 15 billion trips annually, growing at approximately 20% per year, meaning the company is adding around three billion incremental trips each year. The entire global autonomous ecosystem today sits at approximately 50 million trips. Whatever exponential growth curve one assumes, the gap is enormous, and hardware bottlenecks, regulatory friction, and fleet scaling constraints compound the problem.
Between the five- and ten-year marks is where Krishnamurthy sees the real TAM expansion potential materializing, specifically when AV pricing drops below human-driver economics. In the U.S., he pegged the average UberX trip at roughly two dollars per mile, and said AVs need to sustainably price below that threshold before they structurally expand the addressable market. Until then, he expects AV pricing to essentially match human-driver pricing, "maybe plus or minus 10%."
On the bear case that falling AV prices become a race to the bottom on unit economics, Krishnamurthy pushed back firmly. His argument is that lower prices driven by compressing AV cost structures should expand the TAM dramatically — analogous to how ridesharing itself was far larger than the existing taxi market implied. "Even if the margin is lower, which I'm not conceding is the case, the gross profit pool is significantly larger than what you would have had in the other counterfactual world of no AVs existing."
Near-Term AV Partner Roadmap: Zoox, Nuro, Wayve, and the Data Play
While the macro AV timeline is long, Krishnamurthy outlined a near-term deployment calendar that is more active than the market may appreciate. The Zoox partnership, backed by Amazon and operating a vehicle with no steering wheel, is expected to launch commercially on the Uber network in Las Vegas first, followed by Los Angeles. Nuro is on track for a Bay Area commercial deployment by year-end, with early rider testing already underway. Wayve is being explored for London, Tokyo, and potentially U.S. markets. WeRide is already live in Abu Dhabi and Dubai. Pony and Baidu have been announced for further international expansion.
Perhaps more strategically interesting is Uber's own infrastructure play: "Uber Autonomous Solutions," a suite of services designed to support AV partners across data collection, customer support, and insurance. The data collection effort is particularly notable. Uber is deploying sensor-equipped vehicles into its regular driver fleet, targeting two million miles of data collection per month by year-end. With 40 million trips per day globally, Krishnamurthy described the dataset as uniquely capable of capturing edge cases at scale — the exact bottleneck constraining model training across the industry. This is both a service Uber can monetize and a structural advantage that is difficult to replicate.
On balance sheet support for the AV ecosystem, Krishnamurthy described a two-phase approach: active handholding now through equity investments in software partners, OEM offtake agreements, and on-ground infrastructure investment to support launches through roughly 2028, followed by a transition to a fully asset-light model once the ecosystem is self-sustaining. He noted that for every dollar Uber has invested in AV software partners, those partners have raised $2.50 to $3.00 in external capital — a meaningful leverage ratio on Uber's deployment.
The U.S. Mobility Recovery Has Legs — and the Numbers Back It Up
Krishnamurthy provided the clearest explanation to date for why the U.S. Mobility business underperformed international markets for several years, and why the reversal now has conviction behind it. The core issue was insurance: a severe inflationary cycle from 2022 through 2024 forced Uber to raise consumer prices in the U.S., while international markets were largely insulated. Europe and Latin America were still growing 30% year-over-year at a point when the U.S. had decelerated significantly — an inadvertent but clean A/B test on long-run price elasticity.
The unwinding of that dynamic is now well underway. Insurance reform at the policy level began generating relief in 2025, and Uber's philosophy has been to pass those savings directly to consumers rather than retain them as margin. The annual carrier rate renegotiations completed in March 2026 locked in low single-digit increases — "the most benign we have seen in many years," in Krishnamurthy's words. The U.S. business reaccelerated in the second half of 2025, and management is guiding for continued trip and gross bookings acceleration in 2026 versus 2025.
Additional U.S. Mobility drivers include the Wait & Save product, which is showing nearly double the product-market fit in sparse markets versus other product launches, the new Elite premium tier being rolled out across markets, Uber for Business growing at twice the overall Mobility rate, and sparse market penetration delivering double the growth rate of dense markets.
International Delivery: Underappreciated, Not a Black Box
Krishnamurthy pushed back directly on the perception that international Delivery is a low-quality growth story. He described market-leading positions in Canada, the U.K., France, Australia, Taiwan, and Japan, all still growing at 20% or higher despite meaningful penetration levels. Australia, which he called one of the most penetrated delivery markets in the world, is still growing above 30% year-on-year with high margins. Japan has recently accelerated into 20%-plus growth territory. The grocery and retail expansion is contributing meaningfully both as a growth driver and as a new customer acquisition channel — notably, consumers acquired through grocery are converting to restaurant delivery, not just the reverse.
Cross-Platform Engagement: 20% Overlap Is a Floor, Not a Ceiling
Only about 20% of Uber's Mobility and Delivery audience currently engages across both platforms, a figure that Krishnamurthy acknowledged is lower than intuition might suggest. His explanation is that the 20% is largely an organic outcome of building individually strong services, rather than the result of active cross-sell efforts — which have only been pursued systematically for the past one to two years. The newly launched One Search feature, which surfaces delivery options when a user searches for a restaurant on the Mobility app and vice versa, is the most visible manifestation of that active push. In Uber's best-performing markets on this metric, overlap is already above 25%, which he characterized as evidence of meaningful untapped headroom.
Uber One is the other lever: roughly two-thirds of Delivery gross bookings are already coming through the membership program, but only one-third of Mobility gross bookings are. As that Mobility penetration increases, cross-platform overlap should follow mechanically.
AI Productivity: The Hiring Freeze Has a Real Cause
Krishnamurthy offered more granularity on AI-driven productivity than most CFOs are currently willing to provide. As of now, 95% of Uber's engineers are using AI tools on an ongoing basis, and more than 10% of code going into production is autonomously written, with engineers entering only at the review stage. He acknowledged that precise ROI attribution is not yet possible, but said the directional signal is clear enough that Uber has "tempered the pace of hiring" across the company, with the engineering headcount ramp for the remainder of 2026 "significantly lower" than originally planned entering the year. Customer support is the other near-term AI beneficiary, given the enormous ticket surface area created by 15 billion trips annually across a highly heterogeneous global operation.
Financial Framework: $10 Billion FCF, Capital Return Still in the Mix
Krishnamurthy reaffirmed the original three-year framework targets and noted that Uber has been running ahead of all of them. On a trailing three-year basis through Q1 2026, gross bookings compounded at approximately 20% against the mid-to-high-teens target, profits at roughly 50% against the high-thirties to 40% target, and free cash flow conversion came in above 100%. Q2 2026 guidance is for 18% to 22% gross bookings growth with continued margin expansion.
On capital allocation, the CFO was explicit that the priorities — organic reinvestment, AV ecosystem investment, selective M&A, and buybacks — are an "and" strategy, not a forced trade-off. With approximately $10 billion in annual free cash flow, $8 to $9 billion in equity stakes on the balance sheet described as largely financial rather than strategic, a 1x leverage ratio against a stated ceiling of 2x, and investment-grade credit capacity, the company has significant dry powder across all four buckets simultaneously. He was equally clear that the pace of margin expansion will be slower than the historical step-up from lossmaking to profitability, framing this as a deliberate reinvestment choice rather than a structural constraint.