If you took notice of the CES 2024 announcements in the GPU space today and we’re really across the market, you might think that AMD had a bit of a coup in this RX 7600XT announcement.
Unfortunately, you’d be wrong.
In a generation of terrible GPU launches the original RX 7600 was “meh” – it targeted the low end of the market, but with 8GB of VRAM and RTX 4060-like performance, for $20 less, there wasn’t a scenario where someone would give up the additional feature set of the Nvidia card in favour of the AMD offering.
And that’s what the market showed.
Retailers uniformly reported that the RX 7600 didn’t sell and could be had for below MSRP almost everywhere within a month of release.
Nvidia sold some 4060 cards and some of the 4060ti (16GB version) at or near MSRP since their launch – that’s what market share and mind share will do for you.
What AMD Was Trying To Do
AMD is trying to own the entry-level GPU space.
Simple.
The RX7600 XT will compete favourably in the sub-$400 market with the 4060 but comes up well short of the 4060ti 16GB in terms of performance… but it will likely be $100 cheaper.
If your kid plays a bit of Fortnite and Minecraft and you want to get them a PC without breaking the bank, then you might be able to be persuaded that the RX7600 XT is good enough now that it has 16GB of VRAM and that should carry it through for the next 3 years.
And with that, I think AMD has driven a stake through the heart of the RTX 4060 – for $20 more you get double the VRAM and safeguard yourself against growing texture packs we’re seeing in most modern games.
AMD Still Missed The Mark
Nvidia delivered three new GPUs at the mid-range and higher-end of their product set on the same day with the announcement of their Super series cards for the 4080, 4070ti, and 4070.
From a product positioning perspective, this generation of GPUs has been a diabolical mess for Nvidia – they had the original 4080 debacle where they had to “unlaunch” the lower-end 4080 and rebrand it to the 4070ti and the higher-end 4080 has been largely a bust because it was overpriced.
AMD had the opportunity in 2022 and 2023 with this generation of GPUs to really hurt Nvidia.
But they missed badly.
The weirdly named RX 7900XTX flagship came nowhere near the flagship RX 4090 in terms of performance and slotted in against the 4080.
The next cab off the rank was the RX 7900XT, launched at the same time as the RX 7900XTX, and it was slower than both.
The problem was pricing.
| RTX 4090 | RTX 4080 | RX 7900XTX | RX 7900XT | RTX 4070Ti | RTX 4070 | RX 7800XT |
| $1599 | $1199 | $999 | $899 | $799 | $599 | $499 |
Nvidia’s cards have a feature set with DLSS 3, Nvidia Broadcast, and far superior Ray Tracing performance that they are just better cards and the only place that AMD really competes is in rasterization performance.
So logic would dictate that if you want to compete with Nvidia, you have to match their rasterization performance with competing cards and then be significantly cheaper to make the lack of secondary features worth giving up.
AMD didn’t do that. They went for maximum margin releasing slightly inferior cards at slightly lower prices.
Over and over again… while reporting very healthy margins.
They just missed their opportunity.
Nvidia Wins The Day
So we’re right back to where we were as gaming consumers two years ago with Nvidia seemingly in an unassailable position.
In isolation, the 4080 Super announcement is probably what the original 4080 should have been if Nvidia hadn’t been so greedy and fumbled the bag.
We’re getting a card that’s closer to their halo product 4090, at $999.
The 4070 Super puts the AD104 die a little bit closer to full capacity, let’s Nvidia hold their 4070 tier pricing at $599 while running out the lower bin AD104 chips as non-Super 4070 for $549.
The sleeper hit of this announcement will likely be the 4070ti Super (bit of a mouthful).
Using the AD103 die like the 4080, it gets 16GB of VRAM and an expanded memory bus to make use of it.
Frankly, the 4070ti Super is what the 4080 should have been at launch – an $800 card with 16GB of VRAM on a 256-bit bus, in a sub 300W power envelope.
If you’re in the market for a higher-end GPU and you have $800-$1000 to spend, you now have some options… and really, none of them are AMD.
| RX 4080 Super | RX 4070Ti Super | RX 4070 Super |
| $999 | $799 | $599 |
That pricing model and flow is very simple now for Nvidia – there are $200 delineations between their high-end and mid-range consumer cards and the performance difference will likely match the same performance-price variation – the RX 4070ti Super is 20% cheaper than RX 4080 Super, so the 4080 Super will likely be about 20% faster, although the 4070ti Super might be a bit closer at 1080p and 1440p results.
I’m looking to replace my 3070, I use Nvidia Broadcast Voice every day, and so while I would have considered losing that feature a year ago for an $850 RX 7900XTX because of the gaming performance, now it is non-negotiable and I’m debating whether the extra $200 for the 4080 Super will be worth it over the 4070ti Super.
I like AMD products, we run all Ryzen in our house and two of our three GPUs are Radeon cards, but the top end of the AMD product stack is not compelling because the price to features is not good enough.
AMD has lost this generation because of AMD.
Somewhere Jensen Huang is buying more spatulas for his kitchen and laughing all the way to the bank.
