Overview of 20VC: Uber President Andrew MacDonald on Uber’s Growth, Autonomy, China, AI, and the Future of Mobility
Harry Stebbings interviews Andrew MacDonald, Uber’s President and COO, for a wide-ranging behind-the-scenes look at how Uber thinks about execution, product strategy, autonomous vehicles, China, delivery, and AI. MacDonald argues that Uber wins through distribution, disciplined capital allocation, and a long-term focus on consumer value. He also explains why autonomy is existential to the business, why membership is becoming one of Uber’s best growth levers, and how AI is already changing how Uber operates internally.
Uber’s Operating Philosophy
Trust, execution, and “doing the right thing for Uber”
MacDonald says his superpower is not being right all the time, but consistently optimizing for what is best for the company and building trust that people can rely on his judgment.
- He credits long tenure and deep business knowledge, especially in ride-hailing, as a major advantage.
- His approach is rooted in:
- trying to do what is right for Uber,
- being willing to change his mind,
- and knowing the business at a granular level.
Short-term vs. long-term tradeoffs
A major recurring tension inside Uber is whether to prioritize immediate ROI or longer-term strategic bets.
- MacDonald admits he was too short-term oriented at times.
- Dara Khosrowshahi challenged him on this, especially around membership and long-term consumer retention.
- The lesson: sometimes it is better to invest in durable user behavior than in immediate price cuts.
Membership, Pricing, and Consumer Value
Uber One is becoming a major strategic asset
MacDonald now views membership as one of Uber’s most efficient long-term consumer levers.
- Uber One strengthens retention and increases lifetime value.
- Benefits compound over time as members:
- ride more,
- consolidate more spend on Uber,
- and use Uber Eats instead of competitors.
Why pricing still matters
Despite the success of membership, MacDonald emphasizes that price remains central to Uber’s growth.
- To reach far more users, Uber must bring transaction costs down.
- He believes expanding into cheaper modes and products is key:
- trains,
- bikes,
- scooters,
- and eventually autonomous rides.
- The goal is to make Uber useful more often, not just for premium trips.
Autonomy Is “Existential” to Uber
Why self-driving matters
MacDonald says autonomy is existential because it is a better product in many use cases and will improve over time.
He sees autonomy as superior because it can offer:
- lower costs over time,
- better in-car privacy,
- the ability to work, sleep, or relax during the trip,
- and eventually safer transportation.
His core thesis:
Autonomy is as bad as it’s ever going to be today, and every day it gets better.
Uber’s autonomy strategy
Uber is actively investing in autonomous vehicles through:
- equity investments,
- purchase commitments,
- autonomous infrastructure,
- and data collection fleets.
He says this is Uber’s largest standalone investment area.
Why distribution still matters more than technology alone
MacDonald argues that even in autonomy, distribution will be the deciding factor.
- Uber’s global network and 200 million monthly users are a major advantage.
- Autonomous vehicle companies still need utilization and market access.
- He expects AV companies to work with Uber, even if they have their own apps.
How Uber Thinks About the Future of Ride-Hailing
Human drivers vs. robo-taxis
MacDonald is careful not to overstate how quickly autonomy will take over.
Reasons the transition will take time:
- Uber already does roughly 300 million trips a week.
- Human-driven rides are still growing.
- Uber operates in 75 countries.
- Major markets like India and Brazil have fare levels that make human labor hard to beat for years.
His view:
- AVs may represent a small share of trips for a long time,
- but could become very significant in dollar terms in key U.S. cities.
China: The “Wild” Years and the Didi Exit
Competing in China
MacDonald describes Uber’s China battle as one of the most intense experiences of his career.
- Uber was burning enormous sums in subsidies.
- He recalls Uber burning $52 million per week in China during the fight with Didi.
- The company also faced structural disadvantages, including limitations around local platforms like WeChat.
Why Uber exited
He says the decision to exit was pragmatic, not idealistic.
- Uber was not in the lead in autonomy or in China.
- The company needed to focus on turning the core business into a cash-generating machine.
- In hindsight, he says Uber got a better outcome than most Western companies do in China, even though it did not “win.”
Human side of the exit
MacDonald emphasizes the emotional cost of leaving China.
- Many local employees were deeply committed to Uber.
- The company tried to place as many of them as possible into global roles.
Delivery, DoorDash, and the Growth of Uber Eats
Delivery is bigger than many people realize
MacDonald says delivery is an underappreciated part of Uber.
- It is now nearly as large as mobility and has been growing faster.
- He believes the market often undercredits Uber’s delivery business.
Why Uber bought Delivery Hero assets
He explains the logic for acquisitions like Delivery Hero:
- rapid geographic expansion,
- strong local brands,
- better market mindshare,
- and a more compelling combined mobility + delivery proposition.
DoorDash and competition
MacDonald is respectful of DoorDash and says Tony Xu is an excellent founder.
- He avoids rewriting history around whether Uber could have won differently.
- His message: execution matters, competitors are strong, and operating tradeoffs are real.
AI at Uber: Real ROI, But Hard to Measure
The “AI budget” controversy
MacDonald addresses the headlines around Uber blowing through its AI budget quickly.
- He says usage was higher than expected.
- The bigger point is not runaway spend, but that AI adoption is hard to forecast.
Where AI is already working
Uber is seeing concrete value in several internal workflows:
- pricing allocation,
- forecasting,
- marketing QA,
- customer support,
- and process automation.
Examples he gives:
- pricing allocation dropped from 15 hours to 2 hours,
- forecasting from 8 hours to 2 hours,
- marketing QA from 2 weeks to 2 days.
Why ROI is hard to prove cleanly
MacDonald argues that AI value often gets absorbed back into better work rather than headcount reduction.
- Companies get more output, not always fewer employees.
- The cleanest way to capture AI savings may be to set tighter hiring/headcount targets.
Smart budgeting for AI
His preferred model is to combine budgets for people and compute, then let trusted leaders allocate toward the highest ROI.
- This creates flexibility to choose between:
- more engineers,
- more compute,
- or more inference capacity.
Frontier Labs, Aggregators, and the Consumer Interface
Working with OpenAI and others
Uber works with frontier AI labs and is moving from experimentation into implementation.
- MacDonald does not yet see major risks from these partnerships.
- But he understands the concern that companies can feed data to labs that may later compete with them.
Why Uber is harder to displace
He believes Uber is relatively protected because it is a physical-world business.
- Ride-hailing and delivery require local licensing, logistics, and operations.
- He thinks it is unlikely a frontier lab could easily launch and scale a full ride-sharing network.
The UI question: will agents disintermediate Uber?
MacDonald is thoughtful but not alarmist about AI agents replacing the Uber app.
His view:
- if a user says “get me an Uber,” Uber can still fulfill that request;
- the bigger risk is if consumers start from comparison layers and lose loyalty to the Uber app.
He notes that Uber’s transactions are “managed,” not simple checkout flows.
Leadership Lessons from Travis and Dara
What he learned from Travis Kalanick
MacDonald says Travis taught him:
- creative problem solving,
- how to walk into a meeting and change the thinking in minutes,
- and how to explain the reasoning behind decisions.
What he learned from Dara Khosrowshahi
His favorite Dara quote:
“Management comes from an org chart. Leadership comes from the heart.”
MacDonald says Dara leads with low ego, high heart, and a strong willingness to do the hard work himself.
Personal Reflections and Closing Advice
Career advice
The best advice MacDonald says he ever received:
- Say yes to opportunities.
His reasoning:
- you will figure things out,
- you will grow through the challenge,
- and even failure can make you better.
What changed his mind most recently
He says he has become more bullish on:
- humanity,
- longevity,
- and the possibility of solving many diseases over time.
Key Takeaways
- Uber’s moat is distribution, not just technology.
- Membership is one of Uber’s most powerful long-term levers.
- Autonomy is strategically existential, but adoption will be slower and more uneven than hype suggests.
- China was a brutal, capital-intensive battle that taught Uber hard lessons about scale and politics.
- AI is already improving internal productivity, but its ROI is difficult to measure in traditional accounting terms.
- Uber’s future depends on making transportation cheaper, more frequent, and more integrated across modes.