App Economy Insights — Uber: AV Anxiety

Uber's headline sales figure looked like a small disappointment — $14.2 billion, about $70 million under what analysts expected. But that is an accounting illusion. In the UK, Uber changed how it books ride revenue: instead of recording the whole fare and then the driver's cut as a cost ("merchant" model), it now records only its own commission ("agency" model). Same cash, smaller reported number — the switch alone knocked 8 percentage points off the growth rate.

The number that isn't distorted is Gross Bookings — the total value of everything ordered through the app. That grew 24% to a record $58.0 billion, the fourth quarter in a row above 20%. Trips rose 18% to 3.9 billion. The only place growth slowed was Brazil, and for an unusual reason: rivals were bidding for motorbike drivers, so Uber couldn't get enough supply. Demand wasn't the problem. In the US, ride growth actually sped up, because savings on insurance let Uber cut fares — and the cities where fares fell most (San Francisco, Los Angeles) grew the fastest.

Profit is now following volume rather than being sacrificed for it. Core operating profit rose 33% to $2.8 billion, and cash actually generated over the past twelve months passed $10 billion for the first time. That is the money that pays for buybacks, acquisitions and self-driving investment without borrowing.

The argument that decides the share price is robotaxis. The bear case is simple: if cars drive themselves, why does anyone need Uber? Uber's answer is that the hard part isn't the driving software — it's having tens of millions of riders already opening your app, plus the machinery of dispatching cars, cleaning and charging fleets, buying insurance and dealing with regulators in every city. Its evidence: in the cities where robotaxis are furthest along — San Francisco, Los Angeles, Phoenix — Uber's share of the overall ride category is higher than a year ago, not lower.

The market isn't convinced. At about ten times its expected 2027 operating profit, Uber is priced like a business with a problem, not one compounding above 20%. The newsletter's read is that the fear is already in the price and the business keeps improving; the open question is simply how much of today's economics survives when robotaxis are everywhere. The author owns it; analysis, not a recommendation.

Shopify provides the software that lets any business run its own online shop, and takes a slice of what gets sold through it. Sales rose 34% to $3.6 billion, comfortably beating expectations, and the total value of goods sold through its merchants (GMV) rose 32% to $115.6 billion. Cash generation improved sharply: $654 million, or 18 cents of every dollar of revenue, up from 15 cents last quarter.

The reason this quarter mattered is the previous one. Q1 growth slowed slightly and investors began asking whether Shopify's high-growth era was ending. Q2 answered that directly — growth reaccelerated, and management guided the current quarter to roughly 30-plus percent when analysts expected about 27%. That would be six straight quarters above 30%.

The genuinely new information is about AI shopping. As people increasingly ask chatbots to find and buy things, the fear was that AI assistants would go straight to the biggest brands and cut merchants like Shopify's out. The opposite is showing up. Traffic and orders arriving from AI tools both tripled in a year. More importantly, when the AI agent reads Shopify's Catalog — a clean, structured feed of what each shop actually sells, with prices, sizes and stock — the order converts at roughly twice the rate as when the agent scrapes an ordinary web page. Being machine-readable is now worth double.

And the benefit is skewed toward small sellers: 75% of AI-driven orders came from outside Shopify's hundred biggest product categories. Search engines historically buried the long tail; AI assistants appear to surface it. Since the long tail is Shopify's customer base, that reframes AI as a way of getting more customers rather than a threat to them. The author owns it; analysis, not a recommendation.

Nintendo's profits looked spectacular — operating profit up 151% — but a chunk of that is a refund. Roughly $300 million of US import tariffs Nintendo had already paid came back and was credited against costs. That's real money, but it happens once; the underlying business did not suddenly become 2.5 times more profitable.

The hardware story is soft. Switch 2 sold 3.8 million consoles, a third fewer than the original Switch managed in the equivalent launch quarter. That sounds bad, but the installed base is still ahead — 23.7 million machines a year in, versus 17.8 million for the first Switch at the same point.

The interesting part is what people are actually playing. Games for the old Switch sold 34 million copies, up 39%, while Switch 2 games sold only 9.5 million. Because the new console plays the old console's games, the 150-million-strong existing audience keeps buying software even while the hardware generation turns over. In other words, the old library is cushioning the transition — which is exactly what a console maker wants during a changeover, since software carries far better margins than hardware.

Two other quiet improvements: 62% of game sales are now downloads rather than boxes (no manufacturing, no retailer cut), and revenue from licensing Nintendo characters more than doubled after The Super Mario Galaxy Movie passed $1 billion at the box office.

Management left its full-year targets untouched. The real test comes in September, when the Switch 2's price rises to $500 just as holiday buying begins — a price increase into the most important quarter of the year. That, plus the one-off tariff refund inflating this quarter, is why the read here is neutral rather than positive. Analysis, not a recommendation.