« En comparaison d'un PC gamer, les caractéristiques de la PS4 sont voisines d'un CPU bas de gamme et d'un GPU à mi-chemin entre le bas et le moyen de gamme. Si la PS4 est lancée en décembre, celle-ci offrira seulement la moitié des performances d'une GTX 680 (en termes de GFLOPS et textures) qui a été lancée en mars 2012, soit plus d'un an et demi avant.»

Consoles have historically been seen as banner products for the companies that manufacture or design their hardware. Nvidia’s original Xbox win was a sizable windfall for the company in 2002-2003 and the GPU designs of the Xbox 360 and PS3 (built by AMD and Nvidia respectively) were topics of intense discussion when those consoles were new and exciting.
 
Given these upsides, it might seem odd that Nvidia apparently walked away from the negotiating table with Sony on the PS4. It isn’t. While there’s some marketing upside from game enthusiasts, GPU manufacturers don’t get top billing (or typically, any billing whatsoever) on console titles. There’s no lengthy logo crawl or joint marketing. Sony isn’t selling the “Sovidia PS4″, it’s the Sony PS4, period.
 
Then there’s the margin question. Nvidia’s 10-K filing for fiscal year 2003 (that’s calendar year 2002) states that the company made 23% of its total $1.9B revenue, or roughly $400M, on Xbox sales. That’s profits for one year — the original Xbox went on to sell some 24 million consoles worldwide. Nvidia and MS eventually signed an agreement that reduced the cost Microsoft paid for Xbox hardware, but this figure gives us a starting point to work with.
 
Fast forward to the PS3 and its RSX graphics. Two years ago, in January 2011, Nvidia CEO Jen-Hsun Huang told reporters that the Sony-Nvidia deal had earned Nvidia $500M in royalties since 2004. The total number of shipped PS3 consoles by March, 2011 stood at 50 million according to data from the NPD group. Half a billion is nothing to sneeze at, but the numbers imply that NV agreed to much smaller margins on the PS3 than it managed to lock in on the Xbox.
 
One of the most persistent rumors about next-generation consoles like the PS4 is that both Sony and Microsoft wanted to avoid the bloodbath that occurred in 2005-2006. Both companies took three-digit losses on a per-unit basis. Sony’s losses were worse than Microsoft’s, despite the PS3′s higher price tag ($499 and $599 as compared to the Xbox 360′s $299 and $399). Going in to the PS3 launch, Sony executives were hilariously arrogant, with print-ready quotes like “The next generation doesn’t start until we say it does.” When asked about the price, then-CEO Ken Kutaragi memorably declared that the company wanted consumers to think to themselves “I will work more hours to buy one. We want people to feel they want it, irrespective of anything else.”
 
We don’t know when Sony and Nvidia sat down to talk about the PS4, but the Sony’s market position had changed drastically by 2010-2011. The PSP Go was moribund, smartphone sales were rising, and the PlayStation 3 — despite costing far more than Microsoft’s Xbox 360 — had only recently begun to make a profit. Total development and marketing costs on the PS3 from the 2004 Nvidia deal to that time were still deeply in the red.
 
From Sony’s perspective, a number of the gambles it took on Cell, Blu-ray, and on high-margin Nvidia hardware, simply hadn’t paid off. Yes, the system was selling well, and yes, it had boosted the adoption of Blu-ray, but the advent of streaming video and the initial war between Blu-ray and HD-DVD had slowed the new standard’s adoption rate.
 
The PS3 was a massive gamble. It combined multiple cutting-edge technologies, a specialized processor core, an entire PS2 (for hardware-provided backward compatibility), and an integrated wireless solution at the high end. In the long run, that gamble didn’t work. An analysis by Ben Cousins, a 13-year game-industry veteran, writing for Kotaku, shows the profits and losses for Sony’s gaming division from 2004 through the end of 2011. His figures and graphs are available for perusal.
 

 
Converted into US dollars, Sony’s losses stand at $4.5 billion for the PS3 generation. Microsoft, in contrast, has lost about $2.6 billion. These figures run contrary to what a lot of people believe, because Microsoft and Sony have both talked up big years and big quarters — but when you consider the enormous expense of development, the math doesn’t lie. And what did Sony get, in exchange for losing nearly twice what Microsoft has? It got a competitive product. If you look back at what Sony executives were saying in the run up to launch and immediately thereafter, it’s clear that they absolutely expected a crushing victory — not a competitive stalemate.
 
The reason this comes down to margins, not technology, is because Nvidia has been working on its own CPUs for quite some time. No, previous Tegra processors weren’t powerful enough for consoles at this price point (the $99 Ouya is a special case), but NV could likely have whipped up a Tegra 4-class processor in time for the PS4′s launch. It’s easy, meanwhile, to imagine a Kepler-based console GPU — the chip’s excellent performance-per-watt makes such an application a no-brainer.
 
In the end, AMD was probably willing to take a much smaller margin than what Nvidia wanted. Part of that was undoubtedly cost related — AMD was already planning to build custom SoCs that combined console-capable graphics hardware with its own CPUs. Mostly, though, it reflects the relative desperation level of the two companies. Unlike Nvidia, AMD is counting on this semi-custom SoC business to survive. Rory Read has stated he expects embedded semi-custom SoCs (read: consoles and tablets) to account for 20% of AMD’s revenue by the end of the year, up from 5% in 2012.
 
Even after factoring in AMD’s diminished expectations for 2013, a shift that large would reflect hundreds of millions in royalty payments, but it’s predicated on strong tablet sales, strong next-generation console sales, and any remaining payments owed by Microsoft and Sony related to meeting launch windows or other milestones. At this point, with the Wii U reportedly selling far below Nintendo’s expectations and the general x86 tablet market is pretty limp, that prediction looks increasingly optimistic,
 
These macroeconomic trends could hit AMD, and hit it hard. Sunnyvale is depending on strong sales from all three consoles + Temash (next-gen mobile APUs) to drive revenue skyward and help it avoid a shutdown. If console sales don’t spike in the back half of this year or early next, Nvidia may end up having made the smarter choice. That could go double if Project Shield actually takes off — Sony likely wouldn’t have been thrilled with Nvidia attempting to build its own console business, even if it competed in an entirely different business segment.