Is a free speed test website really worth 1.2 billion US dollars just by relying on your internet speed data?

A speed test website, sold for a whopping 1.2 billion US dollars?

Many of you must have experienced this: while you are dueling with other players in a corridor in a game, you suddenly get stuck for no reason. By the time you come to your senses, you have already been kicked back to the game lobby.

After a frustrated post-game review, you still can't figure out whether the problem lies with your network or the game itself?

At this moment, experienced users will usually open this small tool called Speedtest.

Tap the "GO" button in the middle, wait for dozens of seconds, and you will get your upload speed, download speed and Ping value.

No membership required, no subscription needed. Tap to run the test, and that's all its functions offer.

But have you ever wondered how much such a simple website is worth?

Back in March this year, consulting giant Accenture announced that it had acquired Ookla, the company behind Speedtest, for 1.2 billion US dollars.

1.2 billion, in US dollars? Why is a free speed test software worth so much? What exactly did Accenture get for its 1.2 billion dollars?

Why is the speed test code not valuable at all?

From a coding perspective, Speedtest is definitely not a software that seems to be worth 1.2 billion dollars.

Its speed testing principle is very simple. When testing the download speed, the server continuously sends data to your device. When testing the upload speed, your device continuously sends data to the server.

Divide the total amount of transmitted data by the time taken, and you will get the upload and download speeds.

The round-trip time it takes for a message to be sent from your device to the server and for the server to send a response back is the Ping value.

For any device that needs to access the internet, this is the most basic communication function. On Github, a speed test website built based on this principle can be as small as 9KB.

Forcer, the founder of another speed test website Speedcheck, once revealed on a forum that he finished writing the core code for the speed test part in just one weekend.

Nowadays, if you use large language models for vibe coding, it is not difficult to produce a usable speed test tool in just ten minutes.

So, why did Accenture pay the sky-high price of 1.2 billion dollars to buy a website with almost no technical threshold?

What is the purpose of this sky-high acquisition of a speed test software?

In fact, for Accenture, what matters is not the code, but the database behind it.

As a consulting firm, one of Accenture's traditional businesses is to provide enterprises with transformation solutions. After entering the AI era, the main line of this business naturally becomes helping enterprises design new AI workflows.

According to its financial report, Accenture's new AI orders in the first quarter of fiscal year 2026 reached 2.2 billion US dollars, doubling from the same period last year.

As Accenture itself puts it, the most important purpose of acquiring Ookla is to use its database to provide clients in its consulting business with "end-to-end network services required for AI transformation".

This sentence sounds very impressive, but in plain terms, it actually means:

Figuring out how to connect the cloud large language models to the end devices in stores, factories and offices through the intermediate network.

For a simple example, suppose you are the owner of a national chain supermarket, and you want to install AI cameras in your stores to monitor customer flow and shelf conditions.

The most convenient solution is of course to transmit all collected data to the cloud and let the large language model process it uniformly. This is easy to deploy, and you can directly use models with stronger performance.

But in reality, the network conditions in not all stores can meet the requirements of the cloud deployment solution.

The same operator may provide very stable network in the city center, but its service becomes extremely poor in the suburbs with fewer base stations.

If you apply the same solution across the country to transmit all data to the cloud, the result will be that only stores in the city center can operate normally, and the system will become useless once you go outside the third ring road. That means the AI system simply cannot function outside urban areas.

Therefore, before formulating the solution, the enterprise must first find out the network status of each store, then decide which operator to use for each store, and whether the data should be transmitted to the cloud, processed by edge devices, or kept locally for direct processing.

Of course, it is not a difficult task to ask the operators for a list that shows which broadband each store uses and what package it has subscribed to.

But getting these data does not mean you have mastered the actual network conditions.

A gigabit broadband that shows normal operation on the operator's backend, connected to the router through old network cables, and then transmits signals to the mobile phone in the bedroom through load-bearing walls in the living room, may only have a speed of 100 to 200 Mbps left.

The most direct way to bypass this chaotic system is to start from the end device and send a batch of data to the server.

However, for operators, this is actually not an easy task.

A speed test needs to occupy the full capacity of the line as much as possible. If an operator installs a resident speed test background program in the user's device, it is equivalent to making the device continuously upload and download data at full load.

This will not only consume extra traffic, but also greatly increase the burden on the local network. During peak hours, the originally smooth network will be slowed down by this extra load.

This is a definite negative optimization for both cost and user experience.

Therefore, to know the network status of the client, you can only let the user actively run the speed test.

Under this premise, to get results with reference value, you need a sufficient number of people to run repeated tests in different regions, at different times and under different network conditions.

This is exactly where the value of Speedtest lies.

As the absolute GOAT in this niche speed test track, Ookla, the parent company of Speedtest, has accumulated tens of billions of such test records in the 20 years since its establishment.

Every time someone runs a speed test on Speedtest, the backend records more than 1000 pieces of information.

Among these information, besides the upload speed, download speed and latency results, there are also the time, location, operator, network type, and even information about the device itself.

When all these data are aggregated, we will get a network status map covering different regions, operators, devices and time periods.

Then overlay the locations of the stores on this map, and you can roughly judge the network status of each store, so as to select a more appropriate operator and deployment solution.

This is only part of Ookla's intelligence system.

In addition to Speedtest, Ookla also owns Downdetector, which monitors operator service failures, and Ekahau, which identifies signal blind spots.

These sets of tools combined finally generate a network map that extends from the operator's network all the way to the enterprise's indoor Wi-Fi.

Downdetector Explorer

How does the speed test software make money?

However, this network map does not remain valid forever once it is drawn. The network status may change at any time due to adjustments of base stations, broadband lines and operator configurations.

To ensure the map remains valid, you need a steady stream of new data inputs.

Every time a speed test is run, the device and the server will perform a high-intensity round of upload and download. The more users there are, the larger the traffic that needs to be supported.

More than 10 million active users run speed tests on Speedtest every day.

To meet the speed test demands of global users, Speedtest has currently connected more than 15,000 speed test servers around the world.

Moreover, Speedtest relies on its huge user base to keep its data fresh. If it changes the product to a paid one, it will undoubtedly lose a large number of active users.

Although Speedtest also displays banner ads on its webpage, the revenue from these ads is obviously far from enough to cover the server costs.

So, as a company whose core product is a free speed test website, how does Ookla manage to support this huge intelligence network?

In fact, contrary to many people's intuition, Ookla has not lost money in the past 20 years, but has made considerable profits.

In the year before its acquisition, Ookla's annual revenue even exceeded 200 million US dollars.

It does not bear all the operating costs of this network alone. Instead, it adopts a very smart cooperation model: bring in operators, data centers and network service providers, let these companies provide servers and bandwidth, deploy speed test nodes, and join Speedtest's global server network.

These companies are willing to provide servers and bandwidth, and they have their own considerations.

On the one hand, for operators, placing speed test nodes in their own data centers allows users to directly test the speed from their end devices to the operator's internal network. When customer service staff troubleshoot failures, they can run a test from the user's end to determine where the problem lies.

On the other hand, Speedtest can also act as a relatively neutral "referee" between different operators.

Of course, operators can also develop their own speed test websites, but even if they make a website exactly the same as Speedtest, it is difficult for them to get useful results for themselves.

For operators, the most valuable intelligence is to compare themselves with their competitors in the same location, at the same time period and under the same set of standards, to find out in which regions and which links their network performance is inferior to their competitors, and then carry out targeted optimization.

But it is almost impossible to get such information using an operator's own speed test platform.

The reason is very simple: you are the participant, you make the rules, and you are also the one who gives the scores. There is really no reason for people using other operators' networks to run speed tests on your platform.

It's just like if Apple develops a benchmark software and uses it to test Samsung's devices: if the score is high, it shows that Apple's technology is inferior; if the score is low, it looks like Apple can't afford to lose.

No matter what the final test result is, it will not look convincing.

As a relatively objective and independent third party, Ookla can provide data support to every party in the competition without any scruples.

On Ookla's official website, it releases quarterly internet speed reports for different countries and regions. These reports not only show the actual internet speed of each operator in the region, but also include the network quality rankings of different operators in different regions. It even presents awards for the best internet speed performance.