Texas Instruments killed smartphone OMAP about a year after the launch of the Galaxy Nexus. Galaxy Nexus likely used their SoC as an attempt to create more competition in the SoC market. Google tried and failed to do the same thing with Tegra in tablets.
Conversation
NVIDIA didn't drop support for their products early when they left the smartphone and tablet markets though. They pivoted to consoles and kept providing support for the existing products. TI approach was really messed up. They gave up, did a mass layoff + screwed their customers.
1
2
Google ended up having to invest in making their own SoC because they got backed into a corner by the failures of these other companies.
Qualcomm could easily be making much higher end CPUs and providing much longer support for their SoC if they thought it'd be profitable...
2
Qualcomm only seems interested in making higher end cores and providing longer term support for the server market where they can sell expensive high profit margin products themselves. Lack of competition is a major factor. If there were alternatives, they'd be better themselves.
1
1
Yeah, but if Intel can't break them, with years of desktop/server experience and a war chest, who can?
Samsung and Huawei were getting there, but Huawei has obviously had its wings clipped.
1
Qualcomm acquired Nuvia recently and is more than capable of making competitive chips but their focus is the server market. They make very competitive smartphone GPUs already and there was a point when they used to make decent CPUs before they cut their investment in doing that.
1
I think they'll do well in the server market. I don't have much confidence that they'll bring over that architecture for that to smartphones. Qualcomm is also happily providing 10+ years of support for embedded already. They're more than capable of doing it for smartphones...
1
It's a choice not to invest the resources because their customers (phone vendors) largely don't want it and wouldn't take advantage of it. They do provide what most of their customers want. Samsung wants to move fully to Exynos and Google were small fish by themselves though.
2
They're not trying to sell an SoC to end users but rather phone vendors. The only way they would have cared about what Google wants is if Pixels had actually gained a lot of market share. Google's SoC isn't an option for other vendors so I doubt it'll change Qualcomm's ways.
1
Isn't Google's SoC basically a rebranded Samsung? I can see Samsung wanting more of the market, as they already supply screens and others to Qualcomm's other customers, do get a foothold with a decent watch chip and extend out. Can't see Google being interested.
1
It's co-developed with Samsung and isn't simply an Exynos SoC. It's possible that Samsung will get to use it in their devices too. They've had a similar approach with the Pixel Visual Core and Pixel Neural Core which ended up obsoleted by improved Snapdragon SoC capabilities.
Pixel Visual Core was co-developed and manufactured by Intel. Pixel Neural Core was co-developed with Samsung. I think Titan M is based on ARM's secure element designs as baseline with Google doing the rest of the design themselves. opentitan.org will make it RISC-V.
1