Welcome to Master the Meta, the #1 newsletter focused on the business of video games.

There’s no question that streaming -- playing a video game powered by a data center instead of your local device -- is the future of gaming. Streaming is inevitably great for the winning platforms, probably good for publishers (who will sell more games to more people), and a mixed bag for others. One major uncertainty is what it means for the GPU sellers, namely AMD and Nvidia. Let’s figure it out.

The Current Paradigm

The competitive state of GPUs in gaming today is fairly simple. Nvidia and AMD run on the never-ending innovation treadmill, always racing to provide the most powerful and advanced graphics to gamers (and beyond).

In theory, the two companies’ target markets in gaming are the same (PCs & consoles), but in reality they’ve split up the market rather distinctly. AMD powers the Xbox and PlayStation consoles, and holds about 15% market share in the PC gaming realm. Nvidia supplies GPUs for Nintendo’s less powerful Switch console but dominates the PC gaming world.

According to a recent Steam survey, here are the 15 most used GPUs by the platform’s users:

That’s right. Nvidia wins 14 of the top 15 spots. Notably, it hasn’t always been this way, and Nvidia’s most recent and powerful GPUs haven’t made it to the top yet. Impressive, for sure.

The two companies’ margin profiles are also quite different, partially as a result of conquering two different sides of the market. AMD, which made pricing concessions to sell GPUs in volume to Microsoft and Sony, boasts a 40% gross margin, while Nvidia, which sells its GPUs more a la cart to individual gamers looking to modify their PCs, rocks a 60% gross margin.

Objectively, Nvidia is the top dog, superior business, and remains the leader in the technology race. AMD, on the other hand, often wins when companies hunt for “good enough technology at a good enough price.” Although in second place, its technology is nothing to scoff at.

The Future Paradigm

When streaming becomes the norm, walls between various modes of gaming will crumble down. Any game will be playable anywhere with anyone. Your favorite console game will be playable on your phone across town, and you’ll be able to bring your PC games to TVs in other rooms. Hardcore gamers may balk at this assumption, but what it means to be a PC player or a console player will diminish over time.

Naturally, such progress will follow the traditional Innovator’s Dilemma cycle, disrupting from the bottom up. Casual gamers will be the first beneficiaries, while elite gamers who expect the most advanced graphics at the fastest frame rates will take longer to please. Of course, even though it could take several years, technology will eventually improve to the point where everyone converts to streaming. While consoles and fancy PCs aren’t going anywhere anytime soon, it’s increasingly clear that one day they just might.

What does that mean for GPU suppliers? If streaming results in more people playing more games more often (which is likely), then demand for GPUs will rise. That’s great for suppliers. That said, the value chain is destined to change. Individuals will be less likely to buy high-end GPUs because the heavy-duty graphics processing will be taken care of by the newest market participants: data centers. This means that the new buyers — AWS, Azure, Google Cloud, and maybe Tencent — will be fewer yet bigger. Even if demand for GPUs rises, customer concentration risk will increase, and we’ll almost definitely see AMD and Nvidia compete not just on performance… but on price.

While their GPUs will remain strongly in demand (unless certain players try building their own, like Apple did for its mobile devices), Nvidia and AMD will increasingly have to compete on major deals in order to win over the upcoming platform titans. And even if one wins, the pricing concessions could eat into margins. The middle of a value chain is often a tough place to be.

AMD Stays the Course

AMD is used to this business reality; after all, the majority of its gaming business is already reliant on mega-contracts with Microsoft and Sony, as well as the customer concentration and pricing concessions that come with it. In fact, AMD is applying its same strategic playbook to the streaming space. Earlier this year, Google announced that it would partner with AMD on Stadia. From The Verge:

What kind of server matters a whole heck of a lot when it comes to graphical fidelity and keeping the service affordable, given how many players may be using those servers at a given time — and it turns out that Google’s just-announced Stadia cloud gaming service may have struck a balance between power and price by partnering with AMD for a new custom piece of silicon.

According to Google, each Stadia server will contain a custom x86 processor running at 2.7GHz, 16GB of RAM, and most importantly a custom AMD GPU capable of 10.7 teraflops of performance. (They’re running Linux, not Windows, which may matter when Google tries to attract game developers.)

Google wasted no time in comparing that teraflop number against the Xbox and PlayStation competition — where the Xbox One X manages a mere 6.0 teraflops, and the PS4 Pro around 4.2 teraflops.

Good for AMD, and it’s no surprise that these custom chips boast superior performance compared to today’s consoles. While Stadia is no clear winner yet, there’s only a handful of players — including Google — that even have a shot of owning a substantial piece of the streaming pie. Azure will definitely be another key player, and AMD is likely positioned well there, too. After all, Microsoft owns Azure and Xbox, PlayStation will work with Azure for streaming, and AMD already serves both console makers. While nothing is a sure bet, AMD probably feels good about its odds here.

Nvidia Takes a Different Approach

Nvidia, though, traditionally hasn’t competed based on price and has forfeited most of that market to AMD. Despite its continued focus on high-end PCs, Nvidia knows the market will change and is establishing a rather unique yet ambitious streaming strategy.

Instead of appealing to other budding streaming platforms, Nvidia is creating its own: GeForce NOW. GeForce NOW, currently in beta, will let gamers stream hundreds of supported games through their PC, Mac, or SHIELD console (also developed by Nvidia). Like Stadia, gamers must still buy their games individually but no longer need to own advanced hardware to play graphically intense games. Many details like pricing haven’t been unveiled yet.

Whatever the case, this is exactly the wrong decision… for two main reasons.

For one, Nvidia is not going to win. While impressive and beneficial to people like Mac owners, even if its technology is just as good as others’ — which is a tough assumption to make given the scale these platforms must operate at — technology isn’t all that matters.

GeForce NOW might carve out a niche, but its game catalogue (and partners) is incredibly PC-centric, there are no exclusives, there are no direct tie-ins to video platforms (like YouTube and Twitch), and the SHIELD console simply isn’t competitive. If anything, the SHIELD proves that streaming will give rise to multiple new consoles that serve as simple streaming hubs but fail to offer anything new. Xbox and PlayStation have no reason to worry.

Second, even if Nvidia could win, building its own platform is the wrong approach because this isn’t a winner-take-all (or even winner-take-most) market. There will be a handful of big winners, and even if GeForce NOW is one of them, it likely means Nvidia won’t sell GPUs to any of the other players.

Nvidia should take a note out of Netflix’s playbook. In 2007, Netflix was on the verge of launching its own streaming device (dubbed Project Griffin), but Reed Hastings ultimately decided to spin it out. Why? Because Netflix’s success ultimately depended on having as many subscribers as possible and being platform agnostic -- a partner, not a rival -- was key to that. Instead of only bringing in subscribers on the “Netflix Box,” it could also easily sell subscriptions on Apple TV, Amazon’s Fire stick, smart TVs, and more. (Fun fact: Project Griffin evolved into Roku! Both Roku and Netflix are substantially bigger today because of the strategic spin-off many years ago.)

Just like Netflix realized the follies vertical integration and changed course, Nvidia should do the same. Vertical integration might bring around unique benefits, but the resulting scale — of being a rival, not a partner — will be much smaller. Netflix’s key metric is “# of subscriptions” sold; Nvidia’s is “# of GPUs” sold. If Nvidia wishes to sell as many high-end GPUs as possible, then it should once again become platform agnostic and try to become embedded everywhere, much like AMD is doing. After all, if GeForce NOW and the SHIELD fail to win over a large percentage of streaming consumers (and that’s a very real possibility), it’ll be too late. Even pricing concessions are better than not having a seat at the table at all.

What’s going to be a common theme at Master the Meta is that strategy matters as much as, if not more than, technology, yet companies — and potentially Nvidia in the case of streaming — often screw that up.

Generally speaking, when the ultimate metric of success is units sold, vertical integration should be approached with skepticism. Very few companies are optimally positioned to vertically integrate, and very rarely should part suppliers turn into full service providers. I root on the success of GeForce NOW and the SHIELD — the more competition the better — but remain skeptical. AMD is taking the safer approach and it’s already paying off with Stadia. Hope isn’t lost for Nvidia, but the company better hold a few tricks up its sleeve if it wishes to not just take part… but take over.