Cloud Gaming & Negative Latency

Cloud Gaming: The Promise of Negative Latency‍

Cloud gaming is something everyone wants, but it is not a ubiquitous reality yet for a number of reasons, one of which is latency.  

As a quick refresher, the idea of cloud gaming is that the game files are run on cloud servers (AWS, Azure, GCP, etc) and the video from the game is streamed to the player’s device over the internet (you’ll hear people refer to this as pixel streaming). This reduces the processing requirements for the end user (positive) because it has offloaded that processing to the cloud. Yet it increases bandwidth requirements (a challenge) because image data is large and expensive to transfer.  

But Netflix streams just fine, isn’t this problem solved? Unfortunately not. Netflix has a much easier time because they don’t have to transfer in real-time; they cache compressed video content near end users and serve it up whenever it is needed with an added delay so everything appears smooth. Cloud gaming can’t do that. The content cannot be cached because:  

1. It’s being created in real time
2. There isn’t time for lengthy compression algorithms to do their work
3. No delay can be added because every millisecond truly does count

So beyond better and faster compression and delivery of content, what can be done?

This is where Google came up with the idea of “negative latency” for the Stadia cloud gaming platform. They are trying to undercut latency through a few techniques: run-aheads, sending speculative video, and predicting user input.

I won’t go into detail here on each of these techniques because we found a great article, “Exploring Negative Latency”, that explains these techniques in straightforward terms and illustrates the potential impact of each with an illustrative HTML5 canvas game. If you have 10 minutes, I’d highly recommend reading and trying out the example, as it will give you a good idea of what’s possible and what’s not.

‍
Although these negative latency techniques may be somewhat helpful, we’re not particularly bullish on them ultimately solving lag for cloud gaming. They are technically interesting but don’t scale well; often they will increase the compute requirements for cloud gaming and can even degrade the player experience (as is the case with predicting user input). In our opinion, solving for pure latency reduction (better and faster compression and delivery) is the only real solution.

India bans 59 apps from China

In response to the conflict at the border between India and China, the government of India has blocked 59 apps that are owned by Chinese entities. The Indian government has said that these apps "are engaged in activities prejudicial to the sovereignty and integrity of India, defense of India, the security of state and public order."  

This list of apps includes TikTok, Baidu, WeChat, Clash of Kings, UC Browser and Weibo (full list). ByteDance, who own TikTok, expect this ban to cost them ~$6B. According to Sensor Tower, TikTok was downloaded in India 611 million times in 1Q20, equating to 30.3% of its total downloads worldwide in the quarter.

This specific ban is unlikely to last for the long term, yet its near term impact will meaningfully impact the financial performance of many Chinese companies who have significant exposure to India (Alibaba, Tencent, ByteDance, and many others).

‍
Given the macroenvironment right now, this ban between India/China is unlikely to be the last of its kind. This now sets an unfortunate precedent and simultaneously a new risk factor for companies (especially startups) expanding into new markets: “will my app get banned?”  

Streaming platforms growing quickly (7.6B hours watched)

According to Stream Hatchet’s latest Q2 report, Twitch continues to be the dominant player in the streaming space. Just this quarter, Twitch beat its own hours watched record set in Q1 by 62.7% and broke 5B hours watched compared to 3.1B in the previous quarter. What is even more impressive is that the market as a whole generated 96.2% of the second half of 2019’s hours watched in just this previous quarter alone.

From a competitive landscape perspective, Twitch has seen two straight quarters of increased market share (in terms of hours watched) after a dropoff between Q3-Q4 of 2019. YouTube (unlike Mixer, who announced their shut down last week) saw a jump in engagement after a flurry of exclusive signings. They have stayed relatively flat on market share even though total hours watched have been growing.

‍
Takeaway: the live streaming industry is further establishing itself as an important medium of content in the entertainment industry.