Many years ago, when Didi first became popular, plenty of people still regarded it as an illegitimate business. To most, the taxi industry, which required licenses around the world, was territory fenced off by heavy administrative regulation. Traditional taxi drivers were particularly angry: they had paid for qualifications and operated by the rules, only to see another group use a phone to compete for passengers on the same street.
More than a decade later, hailing a car by phone is routine. Taxis have joined platforms too. Experienced drivers either watch for raised hands while waiting for orders on their phones or simply join the private-hire ranks. Once irreconcilable businesses gradually appeared on the same map. Looking back, the old arguments seem distant. Most passengers care little about industry order, may not understand the interests embedded in a license and have no interest in exuberant internet valuations. They simply discover that in the rain they need no longer wait by the road or accept an unreasonable refusal of service. Once, twice, and the habit changes.
Many changes that define an era begin this way.
A couple of days ago, while using a well-known financial-data provider's API, I accidentally retrieved quite a lot of data. Customer service soon called, repeatedly asking what the data were for and which tools I had used, then emphasizing several times that the material could not be used for external publication or other additional purposes. The reminders felt familiar. In the past, the work one person could do at a terminal in a day—researching, downloading data, making tables—was broadly predictable. Now AI can write programs, call interfaces and organize materials. The same person and account have a continuously working capability behind them. The customer is unchanged, but usage boundaries have become sensitive. A business designed around human working speed is beginning to confront machine speed.
At that moment, I felt the anxiety and unease of traditional data providers acutely.
Online discussions often return to Kodak's decline: if we went back to the dawn of digital cameras, how should Kodak have been saved? The answer usually seems obvious: transform earlier, abandon film and embrace digital. From the end of history, foresight always comes easily.
But Kodak saw the future early. The digital camera was born in its own laboratory in 1975. The technologies that later undermined film did not all suddenly arrive from far away.
That is what makes the problem so unsettling. Kodak knew photographs might one day no longer need film. It knew where the future lay. Yet film had to be bought roll by roll, photographs developed print by print. Behind every press of the shutter stood factories, distributors, processing shops, equipment, workers and recurring revenue. Digital photography let people take more pictures without buying film for every shot. The wish to preserve memories grew stronger, while the prosperity built around film lost its foundation.
From inside such a business, one understands the weight of letting go. It means a factory still producing, a channel still profitable, people who have spent years with the company and lives arranged around that income. Assets on the books are often people's pasts as well. So transformation can always wait a little longer. Old products still have customers, old businesses still make money and this year's targets remain unfinished. The new direction matters, of course, but the technology is immature, the market uncertain and investment still needs justification. Every reason sounds reasonable, and another year passes.
Today's internet giants, including Baidu and JD.com, also lack neither talent, technology nor computing power, and are developing their own AI. But an established company must compare any investment with its existing business. For a company like Baidu, search advertising still produces revenue; a way to give users answers faster may also change the route to that revenue. Research needs a budget, businesses need returns and managers answer for present results. New technology carries great expectations, along with a request to disturb the old business as little as possible.
Large companies' hesitation often hides inside these ordinary, reasonable demands.
An entrepreneur facing a market not yet owned can move wholly forward. An established company looks back at what it already possesses with every step. It wants the future too, but hopes that when it arrives, the past's income, standing and order can all remain.
Yet technological progress often changes precisely the conditions on which those things depend.
When information was hard to find, the search gateway had value. When tools were difficult to use, mastery was a skill. When data were hard to obtain and organize, services built around them could charge. AI is shortening many of those steps. Users ask a question and want an answer; assign a task and want someone to complete it. They are increasingly unwilling to care how many pages must be opened, operations learned or software packages used along the way.
We still need to know what happens in the world, but may no longer wish to enter keywords, open a string of pages and piece fragments together. Traditional search must find its place anew within that changing habit.
Data are similar. They remain valuable, and collecting, cleaning and maintaining them still costs money. Researchers, however, may spend less time inside one terminal's interface and instead let their own AI call the data and analyze them. A provider may still be needed; the way customers need it has changed.
The old order has a powerful ability to preserve itself. A successful company knows best how to improve today's business, but today is tomorrow's past. A technology that could change the business itself can easily be assigned to serve existing goals. The company hires people, buys equipment and launches products. Everyone is busy innovating, yet all effort still revolves around old revenue. Past success gradually becomes a restraint. The more one owns, the more people must be persuaded; the more glorious the past, the harder it is to admit that it is losing relevance.
Rowing against the current, failing to advance means falling behind. Attractive accounts do not immediately alert shareholders and executives to changes outside. The old business still earns money, customers do not disappear overnight, and life remains good—or at least defensible. Change happens in small places: someone opens one fewer webpage, cancels a subscription, discovers that work once requiring days can now be done in an evening. These changes are not yet earth-shattering, but they have begun altering future life. For a company, transformation costs are easiest to calculate: equipment, staff, research and marketing can all be itemized. Defending the old business looks much cheaper. The cost of losing the future remains a confused account. By the time the bill is clear, there are often few choices left.
