The phone is not a concept, it's released and people are buying it. Just because a phone isn't available in the US doesn't mean it doesn't exist and doesn't mean the comparison isn't fair. Not everyone lives in the USA.
I'll show you how much the S23U cost:
View attachment 358321
That's my tax invoice for the Galaxy S23U. I got the 512gb(!!!) Galaxy S23 Ultra for $1,589 AUD, the Pixel 8 Pro is $1,699 AUD for the 128gb model. I got a Galaxy Watch 5 thrown in and bought it months before the Pixel 8 Pro.
So if you want to go on price, especially when considering discounts and promotions, the S23U was much cheaper.
The Pixel has poor battery life and reception - this is not a secret for the Pixel line. And especially since Google decided to go high-end with the Pixel 6 series, they should get even more scrutiny with their hardware and since Google moved to Tensor, battery/reception has always been an issue with the phones.
The camera and weight and screen are personal preferences. I've had to turn off Ultra HDR on the P8P as it's way too over the top for me. I find video recording still isn't great on Pixel phones and zoom is just a blow-out win for the Galaxy. Then in addition to that, I'm finding One UI 6 faster, smoother and easier to use with the better notification shade button layout. All personal preferences, but I don't think Google has an advantage here.
What does the Tensor G3 even do? Anything that requires any serious AI gets pushed up into the cloud. I got Video Boost on my phone now and aside how long it takes to process a video, which kinda makes it useless (who wants to go into the settings menu to turn it on, then wait literally hours for the video to process), the results aren't very good. Terribly blown out highlights, way over-saturated colours. The boosted video Pixel 8 Pro kinda puts it on par with the iPhone and Galaxy in terms of brightness, but goes over the top. The Recorder summarisation is surprisingly really bad - it doesn't seem to know what to summarise, gets its sumary wrong and if you have a recording that's too long, it simply won't do a summary at all. Audio eraser changes the sound of the video altogether and makes it sound like you're talking in a tunnel.
The only decent AI thing it does is Magic Editor, which isn't done on-device anyway, it's all done through the cloud and will likely become a feature of Google Photos when you have a subscription, which I already have. So I'm scratching my head as to what the Tensor G3 does really well with AI.
We've had three generations of Tensor processors, I have no expectation that the fourth will be substantially better.