Cloud Gaming: Rethinking the Value Chain

No, You Don’t Need A NVIDIA RTX 3090 To Run Fortnite in 4K 60fps

Fast Scrolling Table of Contents:

• Title
• A tale of two processors — how video games actually work
• Enter Cloud Gaming
• “Fair Winds and Following Seas”: Headwinds (and some Tailwinds)
• Market Deep Dive

In the last two decades of gaming, the arms race between game developers and game computing infrastructure has accelerated. Games with graphics that blur the line between the real and the virtual (looking at you Modern Warfare) demand an inordinate amount of power, technology, and precision that have kept AMD, NVIDIA, Intel, and the like racing to keep up. Gaming is now a $180bn industry [1] with over 2.5 billion worldwide players [2], bigger than movies, music, and streaming.

Competition in gaming is moving to the cloud vs. gaming devices as a whole

Looking back, competition in the industry has largely been about power, portability, and features. The console wars of this millennium between Xbox, PlayStation, and Nintendo (I know adding Nintendo is controversial, don’t @ me) were on the ability of each system to process games the best and the types of games and multiplayers accessible [3]. The console vs. PC conflict of the ~2010s concerned the growing power-to-cost ratio of PCs, declining console exclusivity, modulation of gaming PCs, the rise of powerful consumer graphics cards, and the integration of VR [4]. In the last five years, even gaming laptops have competed with gaming desktops as the former have been able to match power with much higher portability [5].

While the trend of better and more realistic games is unlikely to go away, cloud gaming threatens to upend the way consumers access games. Gaming on a cloud computing network, independent of the ability of the end-user device, is a challenge not just to Microsoft (Xbox) and Sony (PlayStation), but also AMD, NVIDIA, and Intel. Before diving into cloud gaming, I’m going to do a brief aside to cover exactly just how games work.

A tale of two processors — how video games actually work

Video games, at their fundamental form, are deterministic algorithms that computers can run. The game has different states, and the computer takes in the game’s current state, progression of various objects in the game, native commands in the game’s code, and inputs based on the actions of the user and NPCs, to process and bring the game to the next state. In addition to calculating states, the computer also renders the states as graphics and video. Modern games are complex with many states, inputs, objects, and outputs (like detailed graphics) to keep track of, so computers must have powerful components to keep up, chief among them the GPU and CPU (RAM, storage, motherboard, and cooling are also important); these two work together to enable gaming performance [6].

Differences in construction between the CPU and GPU lead to different functionality. Source: A2Z Gyaan

At a high level (CPUs will get a deeper dive in a future post on quantum computing), the CPU (central processing unit) is a microprocessor designed to be the all-around “workhorse” at the center of the computer and meant to act as the hub [7]. The CPU runs all sorts of programs on the computer (Chrome to MatLab) [8] using its large cache and the RAM for low latency and a fetching-decoding-execution process [9]. The CPU notably has fewer cores (physical computing units) than the GPU; even though modern computers make use of multiple threads (virtual execution instructions) through a process called multithreading (sharing resources to make one core function as two) to increase CPU capacity, this is still nowhere near that of GPUs [10]. However, CPU cores are notably more complex and geared towards serial processing, or tasks done in an order rather than at once. CPUs excel at diverse and complex tasks and can be used for internet browsing, managing multiple programs, complex computations, physics engines, etc.

The GPU (graphics processing unit) is a microprocessor just like a CPU, but is specialized for parallel processing [11], the analogue to serial computing where tasks can be run side by side to produce an output. GPUs excel with computations with high throughput that can be broken up into pieces that can be run simultaneously: deep learning, wireframes, 3D images, graphics, bitcoin mining, etc. GPUs have many more cores than CPUs and often require extensive cooling to run [12]. The top-of-the market gaming GPU, the NVIDIA GeForce RTX 3090 [13], has 10,496 cores while the top-of-the-market gaming CPU, the AMD Ryzen 9 5950X (sorry Intel 11900K fans) [14], has 16 cores with 32 threads. This is not to say that GPUs can replace CPUs; GPUs cannot run the complexities of diverse computing operations and are specialized for their purposes.

In gaming, the CPU and GPU work in tandem to run the program efficiently. Notably, lower quality in either component can lead to a bottleneck that handicaps gaming performance [15]. A top shelf GPU cannot run a game in max performance (high pixel density graphics, high frame rate, low lag) without a good CPU and vice versa, though the GPU is more often the bottleneck in practice. In a game like Assassin’s Creed: Odyssey [16], for example, the CPU handles the in-game physics, damage calculations, AI NPC response to players (monsters, enemy assassins, villagers, Athenian/Spartan soldiers, etc.) while the GPU is responsible for rendering each state of the game per the frames per second (fps) and keeps track of frames and objects through the VRAM (video RAM).

A good CPU and GPU could mean the difference between THIS and THIS

Enter Cloud Gaming

Now personally, I feel like “THE CLOUD” is often the go-to response for many in tech as “THE SOLUTION” to make systems scalable. While I don’t feel it’s quite the panacea it’s made out to be (scaling a broken model or just putting tech on the cloud does not a good company make), applying cloud computing to gaming is an elegant solution that is having reverberations throughout the gaming industry.

At a fundamental level, cloud gaming is an innovation in the distribution of gaming (not games themselves, but the process of gaming) to consumers. Instead of each consumer owning a PC or console that has its own CPU, GPU, RAM, cooling, etc. to run high-performance games, computing and storage resources are centralized at a few sources by a company and cloud computing power is “rented out” to consumers to run games. The actual gaming process takes inputs from the consumer’s controller/keyboard, transmits them over the internet to the cloud computing servers, and the game is run on those servers and the video output is streamed back to the consumer. This effectively eliminates the need for “gaming rigs”, requiring only a good internet connection and a display screen (and games, depending on the service) [17]. Cloud gaming allows cloud computing providers to jump into the gaming industry and offer computing power for a lower cost than individual hardware [18] and gain a revenue stream.

Cloud gaming shifts the workflow of gaming more away from the consumer and into shared cloud resources

Cloud gaming has three key advantages:

• Efficient (overall): Consumers don’t always use their personal computing power and even when they do use it it’s not fully or optimally utilized. This creates inefficiencies that can be resolved with fewer, but centralized computing resources.
• Cheaper (for the consumer): By renting more efficient storage and computing, consumers can access the high echelons of gaming performance without having to shell out hundreds if not thousands of dollars upfront [19].
• Wider market (for firms): Cloud gaming brings in users from the mobile gaming market and users who are interested in gaming but cannot stomach the upfront hardware costs.

As the image below shows, the gaming industry is complex with many players, from content creators to game distributors to gaming console/rig creators to chip suppliers (and many other services not shown). Currently, content creators create games that are sold through game distributors to appear onto gaming devices created by firms like Xbox, Playstation, Alienware, etc. These devices also pull in chips from the GPU and CPU manufacturers. Cloud gaming has the potential to disrupt three segments of this value chain:

• Gamer to Gaming Device (Significant): This area is most likely to be disrupted because cloud gaming replaces gaming devices with aggregated servers and cloud computing. Gamers need only a device with a screen and a WiFi connection.
• Chip Suppliers to Gaming Devices (Somewhat): Cloud gaming will still be dependent on good processors to run the games, but the aggregated resources will have more negotiating power and may not need top-shelf processors to run games, instead relying on networks of more numerous, but lower power, processors.
• Game Distributors to Gaming Devices (Somewhat): While this is not the main disruption of cloud gaming, players like Amazon Luna aim to offer games as well as cloud gaming to be one unified platform, potentially disintermediating some traditional distributors.

Cloud gaming has several points of disruption in the gaming industry value chain

“Fair Winds and Following Seas”: Headwinds (and some Tailwinds)

Cloud gaming is a $0.6bn industry and is forecast to grow at a 45% CAGR to $5.4bn by 2026 [20]. This rapid growth is expected to reshape how the video gaming industry operates. The development and proliferation of cloud gaming has several tailwinds propelling it to future dominance, though with a few headwinds that will pose challenges down the road.

Tailwinds

1. 5G and fast Wi-Fi: When games are lo longer local to a device, latency becomes a concern for players due to the user controls (inputs) having to travel to the cloud servers and back as video in order for further controls. 5G can perform up to 100x better than 4G (in reality it is often less than 100x) [21] and internet speeds have gone up nearly 6x from 2010 to 2019 [22], allaying lag concerns and making cloud gaming practically feasible.
2. Size of Games: As graphics and technology get more complex, video games have ballooned in size, increasing ~19x from 2010–2020 [23], with Modern Warfare tipping the scales at 231GB [24]. The cloud storage side of cloud gaming makes lots of economic sense at these sizes.
3. Soaring GPU Prices: Top GPUs are selling for large markups over MSRP (can be double or triple), with the relative supply shortage being driven by pandemic gaming demand, global chip shortages [25], rare earth shortages [26], and demand from crypto mining. Cloud gaming better distributes this cost.
4. New customer sets: Cloud gaming enables deeper gaming penetration of four key growing customer segments: “ultra-casual” gamers who don’t want to fork $1k+ to try gaming, developing country gamers who like to game but cannot afford high quality gaming rigs, e-sports fans who want to participate in gaming, and mobile gamers already accustomed to low-cost gaming.
5. Virtual Reality (VR): The VR gaming industry is $7.9bn growing at a rapid 31% CAGR [27], and the increasing complexity and processing needs of VR [28] demands a better solution that is less burdensome to consumers.

Headwinds

1. Lack of buy-in: High-end gaming rigs are often multi thousand dollar investments, and many gamers are unlikely to abandon their setups until they are obsolete. Other gamers are very focused on lag and ping and may be slow to trust cloud gaming. These can create barriers to adoption that can prevent early scaling of the industry and have network effects of preventing further adopters.
2. Re-fragmentation: Like the story with Netflix, incumbents (like Disney) that are disrupted will rarely “go quietly into that good night”. As we will cover in the next section, firms like Xbox, PlayStation, etc. are moving into cloud gaming and could bring back an era of exclusives that might fragment the cloud gaming market and force consumers to own computing subscriptions for different types of games. We’re starting to see some of this with Microsoft pulling its games from Boosteroid’s platform [29].

Market Deep Dive

Cloud computing being defined and the major factors being identified, let’s dive into the specifics of the market players. Who are the cloud gaming firms, what do they offer, and which ones are most poised for success? For this article, I am only looking at the B2C cloud gaming companies, not the B2B ones that offer cloud gaming infrastructure to other firms. There are 20 firms I have identified in the B2C cloud gaming space (details below). This table shows us the variety in firm types, business models, geographies, etc. [30–49] There are 5 key takeaways from the below analysis:

• Monetization Strategy: Most firms are charging a monthly subscription fee, but some rent out server capacity on an hourly basis. Many firms have “freemium” models to give users some capacity or some games and let them try. The hourly rental model seems to be better for firms just renting out servers for processing to low-use gamers, but not offering games.
• Bundled Business Model: Some firms are BYOG (Bring Your Own Game), meaning that they only offer processing capacity but users need to own and connect games to the platform. Other firms are CDGG (Cross-Device Gaming + Games), meaning that they offer games as well as game processing as part of the subscription.
• New Entrants: Only 4 of the 20 are gaming incumbents, with the majority of companies being new to gaming. The incumbents are likely moving into this space to avoid losing business.
• Globalization: Cloud gaming is not just a US phenomenon, with firms like Abya, Nware, The Gaming Project, and others focusing on South America, Europe, and India.
• User Growth: While the overall gaming industry is larger, cloud gaming firms have cultivated a substantial user base, with many in the millions of users.

Depending on your computer some of the gridlines in this table might appear missing

Two factors that are especially important are the bundled business model and the new entrants. In the below graphic, I organize the industry into a landscape matrix on these two factors. Doing this lets us break (the four quadrants) and find more direct competitors. Q1 is firms new to the industry offering CPGG; Q2 is incumbent firms also going CPGG; Q3 is firms new to the industry doing BYOG; Q4 is incumbent firms doing BYOG.

• The list of incumbent firms is small, but this is because many incumbent firms are doing B2B cloud models (Ubisoft) offering services to other B2C firms. Incumbents are making moves into cloud gaming to avoid losing critical revenue streams through the industry shift. However, for players like EA, Xbox, etc., this means offering less broad titles.
• The industry as a whole is moving towards CPGG, and I expect the most successful firms will be here. BYOG was a good starting-off point for the industry because it was lower cost, but as users have grown, it makes more sense to offer a consolidated gaming product to get more to switch and to give cost savings by disintermediating the middlemen.

High Conviction Firms

These are the firms I have high conviction in for the future in terms of success/growth:

In summary, cloud gaming is a busy space that is only heating up as time goes on and firms and consumer adoption threatens to upend the game process and some of the middlemen in the gaming value chain. The industry, once mainly small firms offering processing power and one that has seen ups and downs before [50], has evolved with larger incumbents (Xbox) and large firms (Amazon) entering the market and the service model moving to CDGG. Internet speeds, game sizes, processor prices, new markets, and VR are working in tandem to propel cloud gaming onto the mainstream, and users have grown to the tens of millions globally. The future of the industry is predicated by consumer adoption and new entrants. Strong consumer adoption will continue to drive scale in cloud gaming, while a major risk is incumbents entering the market and re-fragmenting access to gaming (akin to Disney+ and Netflix). Cloud gaming is certainly not a trend that is going away anytime soon, but the question is how “open” gaming will remain.