The Reuters agency has now come up with interesting news. According to her information, it has been clear for about two years to whom Sony has entrusted the development of the hardware – that is, the SoC with the processor and GPU – of the future PlayStation 6 console, which is still a few years away. The selection appears to have ended without changes against current practice. Following the PS4 and PS5, as well as its now-revealed PS5 Pro upgrade, the PlayStation 6 will be based on an AMD chip, so it will have GPU technology based on Radeon graphics.
This choice is logical for Sony, as it should mean seamless software compatibility with games for the previous two generations of consoles, and once again ease development for multi-platform game studios used to targeting PCs. That is, especially if the x86 CPU architecture will be used again with Radeon graphics, which in theory may not apply, but it would make sense. So this piece of information has value mainly as a confirmation that Sony has not proceeded with an unexpected change of supplier.
But more interesting is the information about who Sony did not choose, although it was interested in producing hardware for the PS6 and AMD taking over this business: namely Intel. According to Reuters, he applied for the contract and was an opponent of AMD in the final decision. This is said to have already taken place in 2022, when Sony opted for an existing partner and Intel refused. So AMD’s participation has been “locked in” for some time now.
According to Reuters, Sony was probably seriously interested, but the cooperation with Intel failed because the company was not willing to go with the margin as low as Sony wanted, or simply because AMD offered an even lower price. Care must be taken with interpretations. This is not to say that Intel had any big eyes or was too greedy – the company even reacted atypically to this report by objecting to its tone.
But it’s possible that Intel’s proposed solution had higher production costs (due to a worse ratio of graphics performance per silicon area) than AMD’s, and the margin problem really stemmed from that. Or Sony took into account that AMD’s solution is more proven and Intel would have to lower the price to balance this.
Anyway, Intel failed to get the PlayStation 6 as a “design win” for its GPU technologies. At the same time, such a partnership could have been a good way to strengthen Intel’s position with game developers and get its GPUs more respectability and a better degree of optimization in games.
It would help Intel the most if the SoC was manufactured in its factories
According to Reuters, the processor (with an integrated GPU) for the PlayStation 6 may be a deal that will bring cumulative sales of up to $30 billion to the “winner”. That sounds like a lot, but consoles are a low-margin business, so those sales would be relatively low-margin, although even an increase in sales would help Intel’s numbers now. But losing this option will hurt Intel elsewhere.
The possibility that the processor designed by him would then also be produced in his factories was probably very interesting for him. As we’ve discussed in several articles dealing with Intel’s current financial and factory construction issues, Intel is in desperate need of outside clients to help bring the economic scale of its chip manufacturing closer to that of TSMC, allowing Intel to compete with the leader and finance operations. and the construction of their factories.
While the $30 billion for PS6 chips wouldn’t be a huge net profit, a lot of that money would stay in his factories if the chips were made on his process (apparently 18A technology). Intel Foundry Services would thus improve its revenue by many billions and at least mitigate its loss. In other words, such a large order would do a lot to keep factories busy (thousands of wafers per month, according to Reuters). And also the example of Sony would serve as a reference and could convince other customers.
Wafer produced by Intel 18A process
So now Intel has to look elsewhere for external customers. It is not entirely without success, for example, it has now been announced that Amazon should manufacture its AI chips on its 1.8nm process. But we are probably still far from the state when Intel would have a “full diary” and in its pocket orders ensuring enough work and funds for its long-term plan for the development and operation of its silicon factories.
Broadcom tried it too. Nvidia doesn’t?
The interesting thing is that according to reports from Reuters, there was another company bidding for this contract. Oddly enough, it wasn’t ARM, where you might expect it, but Broadcom (or the former Avago, which Broadcom swallowed and then took its name). At the same time, the latter has nothing to do with consoles and gaming (apart from the fact that it can provide wireless connectivity for consoles). It is likely that Broadcom was offering Sony an SoC design with cores licensed from ARM, and the GPU as a core component was likely to come from that. In that case, the console would run on a powerful GPU configuration derived from the Mali and Immortalis line of graphics used in mobiles.
It must be said that this would be a rather untested solution, because so far no one has tried to “scale up” ARM architecture GPUs to the level of higher discrete GPUs. Also, the software layer and middleware for game developers would probably be a lot more immature and difficult than with GPUs from AMD or Intel. According to Reuters, Broadcom also fell out of the selection relatively soon, in the later stages Sony was only deciding between Intel and AMD.
There is no mention that Nvidia is bidding for the contract or that it will advance to the later stages of selection. If it’s not just that the sources aren’t familiar enough, it could be that the company simply wasn’t willing to offer Sony anything at a price low enough to negotiate with. According to some rumors, Sony was also not satisfied with the last cooperation it tried with Nvidia (it was the original Xbox console and PlayStation 3 at Sony), just like Microsoft, and it is possible that this is still an obstacle in the cooperation of the two companies.
Theoretically, there is a possibility if, for example, Broadcom did not enter the competition with the fact that the GPU in its SoC was just licensed from Nvidia, which would certainly be a solution with a better chance of working well. However, we would rather expect Nvidia to submit an offer on its own.
Source: Reuters
Source: www.cnews.cz
