Technology is a game of 1 billion players

In the spacious workshop, rows of parts and components with the Mihoyo logo are carefully transported into the assembly line and ready to go.

Soon, they will go through a series of delicate processes: sorting, routing and assembly, loading them into the game, and lighting them up in a computer room to become part of miHoYo's cloud gaming world.

For many players, cloud gaming has an inexplicable attraction, and this gravitational pull does not come from the game, but more from the technology itself, which has a sense of the future that it cannot aspire to.

After all, streaming the rendered game to the local device in real time through the high-speed network means that we will no longer need to install it, nor do we need to consider the computing power configuration of the local hardware, as long as we have the network, we can play everything.

The distribution attributes of cloud games are also in line with the expectations of almost all game manufacturers - game works can get rid of the threshold of devices and channels to the greatest extent, and directly reach the largest number of players.

But technology never comes out of nowhere. On the occasion of the official launch of the cloud gaming version of Honkai: Star Railway, miHoYo also released a documentary about the company's long exploration and practice in the field of cloud gaming.

It's also a story of technology and dreams.

I remember my first experience with cloud gaming in 2017 with Final Fantasy 13 Cloud Gaming Edition from G-Cluster, a Japanese cloud gaming service.

It's just that the mosaic pictures and stuttering frames brought about by the severe network delay at that time really didn't leave a good impression on me, and cloud gaming is more like a beautiful concept, which can be viewed from a distance and cannot be blasphemed.

Today, seven years later, you can find a thousand-yuan Android machine on the market, open the browser that comes with it, and log in to "Cloud Star Railway", you can enjoy the high-quality modeling and light and shadow effects that are no less than those on the PC side.

This directly broke the shackles of the concept that whether the game experience is good or not depends on the hardware configuration, and pushed the door to a new world.

Ideally, cloud gaming means that anyone can play faster and more conveniently on all devices, which is completely different from how we normally play games.

No need for **, does not occupy storage space, liberates hardware computing power, and allows players to experience high-quality games with low burden, which is the most direct attraction of cloud gaming.

For miHoYo, helping players break through the computing power limitations of local hardware is one of the important values of cloud gaming.

Behind every player who is playing, there is a server computer carrying the weight forward, and we are like opening a 24-hour Internet café.

However, as the project progressed, some pain points were also highlighted.

First of all, there were not many servers available for miHoYo on the market at that time, and high-performance public cloud servers that could meet the standard could serve more AI needs.

This also forced miHoYo's cloud gaming team to optimize the underlying layer of the game so that it could run on most public cloud servers.

But then comes the second problem – the high cost of renting servers.

In miHoYo's view, the public cloud server pursues large and complete functions, which is not only difficult to fully adapt to the game, but also brings a waste of computing power.

These redundant things make the server cost very high, and it will also lead to high expenses for players, which is not what we want to see.

To sum up, for miHoYo who wants to take their cloud gaming experience to the next level, the public cloud is only available, but it is not easy to use.

It's like buying a suit, ready-to-wear as a universal solution, as long as you don't have a lot of sizing details, you can indeed buy a ready-made set to take home.

However, in contrast, the tailor-made customized model is selected for the buyer's service in terms of version and fabric, which can better meet the personalized needs.

After weighing it again and again, it became miHoYo's only choice to go to self-research by himself.

The goal of the miHoYo cloud gaming team is clear, that is, to create a set of cloud server hardware solutions that meet the demand for computing power, are stable, reliable, can be deployed on a large scale, and can be profitable.

While we don't necessarily know the most about servers, there's no doubt that we know our games best.

In the past, Amazon, Facebook, Microsoft and other tech giants have tried to take advantage of their large data processing centers to create cloud gaming systems.

But many failures, including Google's Stadia, show that cloud gaming platforms are not as simple as they seem. Some of the obstacles come from technology, while others are blamed on the lack of a sufficient number of users, and ultimately the business model is unsustainable.

There's a metaphor in the industry that says cloud gaming is like a meat grinder for resources and money. In addition to the technical issues, the bigger challenge is how to find a balance between the quality of the gaming experience and the cost of hardware operation, which is the key to sustainable development.

miHoYo is no exception, on the one hand, self-built servers mean that there are more choices in hardware, but on the other hand, due to the lack of a universal solution, it is bound to face the first-chain problems that are common in the real industry, such as the delay in manufacturing by manufacturers and the insufficient supply of unpopular accessories.

After studying the performance consumption of its own games and clarifying the product form and functions, the miHoYo team found their own breakthrough.

Similar to mobile phones and PCs, servers are also composed of a number of key components, basically, except for the CPU, GPU, hard disk, memory and a few large items, as long as the parts that are not used well, they are listed in their self-developed list one by one.

For example, in order to solve the remote management function, the miHoYo team developed a general-purpose BMC board called Milo. It is like the "big butler" of the server, which not only allows the operation and maintenance personnel to achieve a higher degree of freedom in operation and maintenance, but also expands the selection range of motherboards by miHoYo, so that they no longer have to rely on expensive parts from a single manufacturer to reduce costs.

In terms of power management, the power supply used by miHoYo was not only poor and expensive, but also very simple, so they also redesigned the power supply scheme, developed their own power conversion board, and adjusted the power supply of the whole machine to the power supply of the chassis, which has been improved in terms of cost and reliability.

After the hardware selection was confirmed, miHoYo's technical nerds even customized a chassis so that the self-developed parts could be placed in the standard server cabinet, effectively increasing the density of the server.

However, it is not enough to design a complete set of solutions, and in order for cloud servers to mature, they must also realize the verification of the production process, and finally move towards mass production and form a scale.

Unlike assembling a server in a lab, mass production in a factory with thousands of machines is another difficulty.

After the game is installed, the machines will be placed in the high-temperature aging laboratory, subjected to a 48-hour stress test, and finally placed into a box printed with the miHoYo logo, before completing the entire production process and waiting for shipment.

At this point, the problem on the hardware side is solved, but what about the experience side?

For players, whether cloud gaming can provide an experience close to that of local games is an important reason why they are willing to use cloud gaming, especially the lag and screen tearing caused by network latency, which is also an objective uncertainty of the cloud gaming experience.

With today's technology, it is almost impossible to completely eliminate latency, and in order to minimize this uncertainty to an acceptable range for users, miHoYo can only choose to increase the number of servers to improve the connection experience for players.

This involves the large-scale laying of the computer room, which naturally requires a huge investment of money.

In China, miHoYo has built dedicated cloud server rooms in many cities in the north and south, and the service scope is enough to cover the whole country.

The servers followed the trucks here, split them and put them on the shelves one after another, and then passed the software running test verification, and finally they will be lit up to work, and in the next few years, they will continue to provide stable cloud gaming services to the surrounding areas.

For the exploration of unknown areas, they will continue to continue, a huge cloud game territory, is slowly unfolding.

This is not the first time that miHoYo has predicted cutting-edge technology.

In the documentary "Chronicle of Chinese Games", Liu Wei, the president of miHoYo, also known as Viagra, revealed that in January 2017, when the "Genshin Impact" project was established, even the best performing iPhone 7 mobile phone at that time could hardly meet the standard of running an open-world mobile game.

To put it simply, at that time, miHoYo was already designing games for mobile terminals 2-3 years later, and its confidence came from the deduction of the development of computing power

I think it's a big bet for us, but it's an option for our company.

Now in the layout of cloud gaming, miHoYo has once again stood at the forefront of technological prediction. Perhaps in 3-5 years, cloud gaming will become the main way to play games The goal is still difficult to achieve, but whether it is an objective trend or a subjective perception, we have to admit that cloud technology will occupy a place in the future.

In fact, none of the world's largest cloud computing service providers, Amazon, Microsoft, and Google, started out with cloud services, but these three companies took 65% of the $500 billion market. This also means that cloud gaming is a huge market that all the giants are eyeing up, and there is a huge platform opportunity lurking.

At a time when gaas is all the rage, cloud gaming is also the form that best suits its service spirit - using the most advanced technology to serve the widest audience.

But is cloud gaming the ultimate goal of miHoYo?

Looking back on the past two or three years, you will find that miHoYo's focus on hard technology has long gone beyond the scope of ordinary game companies - whether it is investing in brain-computer interfaces, fusion energy research, or co-development of rocket companies, it shows their pursuit of new technologies and cutting-edge science.

As well as Apple's latest Vision Pro headset, when most companies are still waiting and watching the era of spatial computing, miHoYo has chosen to embrace the opportunities brought by new technologies - "Honkai: Star Dome Railway" has become the first batch of games officially announced to be launched on VisionOS, which is quite a sense of laying out the virtual world in advance.

Based on this consideration, I suddenly remembered a grand and fascinating vision that miHoYo had mentioned two years ago:

By 2030, create a virtual world in which one billion people around the world want to live.

This may also be a spark of fire, when a few years later, when we take out our portable devices, click on the browser, and play a hearty game, we can't help but look at the sky and remember a cloud called miHoYo.

Dreams still have to be had, what if they come true?