One of the difficulties with reviews is trying to determine the overall context, as well as looking toward the future. The easiest thing to change on a graphics card is the price, but the market will ultimately determine where any product falls. If it's overpriced, you can expect that to trend downward over time. If there's a ton of demand, you can expect prices to trend upward. So, I have to look at the everything the card offers, and then factor in pricing as the final point, understanding that MSRP has become very disconnected from retail prices.
In terms of performance, the 4070 Ti is a card that matches the 3090 almost exactly, at a lower price, lower power, with some extra new features. There are cases where it's definitely slower, and cases where it's clearly faster. 3090 performance for $800 isn't the end of the world, which ultimately leads me to the 7/10 score. Maybe it could have been 6/10, but basically performance and features is in the 8 or 9 range, and then price adjusts that.
Rarely do graphics cards outright fail to provide anything useful. In recent history, the GTX 1630 is as close as we've come to a pointless product. It actually made the RX 6400 seem slightly better! But just because everyone is shouting that the 4070 Ti (and 4080) are overpriced doesn't mean a thing if the cards are selling. We don't have data on how fast they're selling, or how many cards there are. We also don't have clear data on the BOM — bill of materials — not to mention the R&D costs. However, I do know that Nvidia (and AMD and Intel) have been at this for a long time, and they're not stupid.
The reason they've chosen the $799 price point is because that's what the number crunchers think the market will bear while delivering optimal returns. If reality diverges from that, the price will start to fall, but I'd be surprised to see more than a $100 adjustment over the life of the product. If the price is really off, Nvidia will ultimately discontinue the part and create a new model at a different price to compensate, rather than "admit" that MSRP was totally wrong.
And people do need to look beyond the model number. What we have is basically Nvidia deciding that it will stretch out the range of performance offered. Where we used to have 50, 60, 70, and 80 models, now we have 50 through 90 models, plus most likely Ti variants on each (eventually). And rather than a 5–10% performance bump between the top models, we're now seeing a gap of 20–30%. That's not all bad.
The 3090 being 10% faster than the 3080 at launch while theoretically costing twice as much always felt stupid to me. Extreme Edition Intel CPUs that were 5–10% faster than the parts that cost half as much also felt silly. But both of those things happened, and apparently enough demand exists that rather than getting rid of the extreme parts, they're becoming more normalized, and pulling prices up on the high-end and midrange models.
For now, this is the reality the big corporations are trying to create, and the only way that reality fails is if consumers and businesses reject it and refuse to buy the parts. Given the RTX 4090 is still selling for over $2,000, it would appear there are enough people ready to fork over large amounts of cash that the strategy is working.
Facebook
Twitter
Instagram