Uber's budget broke before the product benefit did
Uber burned through its entire 2026 AI coding-tools budget in four months, but that is not the core issue. The bigger question is whether the company is paying more for an AI stack that may make engineers more active without making the product measurably better for consumers. Management's own operations chief said it is still hard to connect higher AI use to more consumer-facing value, saying the link to "25% more useful consumer features" is "not there yet". In that sense, higher AI spend looks hard to justify if it is not delivering proportional value.
That is especially important because Uber's underlying business still looks healthy. In the first quarter, trips grew 20%, gross bookings rose 21% on a constant-currency basis, and non-GAAP operating income increased 42%. The company also returned $3 billion to shareholders in the quarter. So this is not a broken platform. It is a strong platform dealing with a new, poorly understood cost layer.
The investment tension is clear: if AI usage keeps rising but the consumer experience does not improve in an obvious way, investors may be funding higher costs rather than better retention, usage, or pricing power.
Uber's AI problem is governance and proof, not adoption
Adoption was never the issue. Engineers picked up AI tools quickly, and management encouraged that behavior with an internal leaderboard ranking teams by total AI tool usage. The problem is what happens when that kind of adoption meets a usage-based pricing model.
Token-based billing turns AI spending into a variable cost
With traditional software, companies mostly bought seats and could forecast the bill. With AI coding tools, every time an engineer opens their laptop, the meter runs. That makes costs much harder to pin down in advance.
Uber's CTO said adoption became so deep that the company could not predict the cost. Power users can run ten-plus parallel worktrees. Once engineers are leaning on the tool, usage can expand in ways that licenses never did.
That helps explain why the 2026 budget disappeared so quickly. This looks less like a simple discipline failure than a budgeting mismatch: treating a utility-like expense like a fixed software contract.
A leaderboard can reward effort without rewarding outcomes
Measuring success by tool usage can create the wrong incentives. Usage says how much AI is being called on; it does not say whether that work improves the rider experience, reduces failures, or ships better product.
That is the real watchpoint for investors. Broad adoption shows the tool is popular and embedded. It does not, by itself, show that the spending is producing commensurate business value.
What matters for Uber investors now: platform first, AI second
The cleaner lens here is platform first, AI second. Uber's core business does not need a clean AI success story to remain interesting. What matters now is whether this spending wave eventually produces a better customer product or a more efficient operating model.
Management has also framed autonomy as existential for Uber's future, on a timeline spanning years rather than months. That keeps the broader long-term story intact even while the near-term AI ROI case still looks incomplete.
Why the bull case still exists
The bullish argument is not that Uber is a neat AI-coder story. It is that the company may be early in building a broader efficiency and automation moat. If that longer-term autonomy and commerce stack works, today's messy AI spending could look like an expensive learning phase rather than the final answer.
Why the bear case still matters
The bearish argument is about capital discipline. Uber has already said the link to "25% more useful consumer features" is "not there yet", and that higher AI usage is not delivering proportional value. At the same time, the company returned $3 billion to shareholders in the first quarter. If AI costs keep rising while the consumer payoff stays vague, investors are effectively subsidizing a new expense layer.
What to watch over the next two quarters
- Consumer impact: Are riders and other users seeing a cleaner, faster, more useful product?
- Cost controls: Does AI spending start to look managed instead of surprising?
- Engineering outcomes: Are gains showing up in releases, quality, or deployment efficiency?
- Management framing: Has the conversation moved from budget shock toward governance and measurable ROI?
For now, the constructive view on Uber still rests on platform strength and cash generation. The AI story becomes more compelling only if management can show that higher spending is producing either a better product or a cheaper way to run the business.