There are three major classes of products that AMD announced. None are available yet, but they're all coming soon.
The most important is the Ryzen Mobile 4000 series. AMD has already launched Zen 2 based CPU parts for mainstream desktops, servers, and HEDT, and soundly trounced Intel in most of the ways that matter in all of those markets. Conspicuously missing was laptops, where AMD should be able to offer a terrific product because their CPU cores are so much more efficient than Intel's. So we knew that the Ryzen Mobile 4000 series was coming, and even had a pretty good guess of what it would be called.
Today, AMD announced that it will offer 8 CPU cores (16 threads), a GCN-based GPU with 8 compute units, and a CPU max turbo of 4.2 GHz. Also, that's a 15 W part. AMD is also offering a 45 W version of the same thing, which laptop vendors are apparently only interested in using for gaming laptops. Naturally, there will be cut-down 6-core and 4-core versions of the same thing, and with analogously fewer GPU compute units.
It's interesting that AMD is finally offering an 8-core APU. They had topped out at 4 cores ever since Llano way back in 2011. AMD has long been known for offering more cores than Intel, even in the days when AMD's CPU cores weren't very good. So it's interesting that their APUs stayed at four cores for so long even as Intel moved to six and then eight. Well, AMD has sold plenty of 8-core APUs for consoles, but low power Jaguar cores are a different beast entirely.
At multi-threaded CPU or integrated GPU performance, there's little doubt that AMD will handily crush anything Intel has to offer in the same power envelope. We've already seen the same CPUs compared in desktops, and Intel's integrated GPUs aren't very good.
But there are two areas where Intel may be able to hold onto a lead. First is single-threaded CPU performance. There's quite a gap between AMD's max turbo of 4.2 GHz in a Ryzen 7 4800H and Intel's 5.0 GHz in a Core i9-9980HK--and that's much larger than the desktop part gap between AMD's 4.7 GHz and Intel's 5.0 GHz. A 15 W power envelope isn't going to clock terribly close to 5 GHz for very long, so AMD might well commonly catch Intel in single-threaded CPU performance there. But when a single CPU core has most of that 45 W all to itself, that allows a pretty high clock speed. You can't push several cores close to 5 GHz and stay inside of 45 W, so the advantage will eventually shift back to AMD as you start using more CPU cores.
The second is low idle power consumption, and hence long battery life. AMD has never been competitive with Intel there since Conroe dropped way back in 2006. There have been a variety of reasons for this. One is the lack of LPDDR* support, which the Ryzen Mobile 4000 series finally adds (LPDDR4X). But that hasn't been the only reason, and we'll see if AMD managed to catch or pass Intel there. Of course, if an Intel-based laptop can get you 10 hours of battery life when idle at desktop, and an analogous AMD-based part can "only" get you 9 hours, do you really care about that difference? If the battery will last longer than you need it, that's good enough.
AMD customarily doesn't publish prices on laptop parts. The laptops themselves will launch when they're ready. Laptops have to do enough custom stuff that to the extent that laptop CPUs even have a launch date, it isn't accompanied by a whole lot of laptop launches. But AMD does expect the first laptops to arrive in the first quarter of this year.
The second major announcement was the Radeon RX 5600 XT. This takes the same Vega 10 GPU chip as in the Radeon RX 5700 XT and cuts it down further. It doesn't just disable compute units and clock lower, but also disables a memory controller. This will be AMD's first memory size that isn't a power of 2 since the Radeon HD 7970 had 3 GB way back in 2012. For that matter, it will only be the second time that AMD has gone with a non-power of 2 memory capacity since buying ATI way back in 2006.
As the Radeon RX 5600 XT is just a new bin of an old part, there isn't really that much to know about it. It's going to be faster than an RX 5500 XT and slower than an RX 5700, and with a price tag of $280. Exactly how interesting of a part that is depends on exactly how fast it is. There won't be reference boards, but only custom boards from AMD's usual board partners. The new cards go on sale on January 21.
The third major announcement is the final third generation Threadripper part, the Threadripper 3990X. AMD had previously said that a 64-core part was coming this year. Now we have a date: February 7. We also have a price, where it will be one of the rare parts whose price tag is actually in the name: $3990. Normally, I'd call that $4000, but since it's in the part name, I'm willing to call it $3990.
While the Threadripper 3960X and 3970X could justifiably be called legitimate gaming CPUs that have a bunch of extra stuff added (more cores, more memory, more PCI Express connectivity, etc.), that's not the case for the 3990X. It will have the same 280 W TDP as the parts with fewer cores, but necessarily have to eat up more of that TDP on powering extra cores and bandwidth to connect them all. That means it will clock lower, with a max turbo of only 4.3 GHz.
The Threadripper 3990X will be a rather dumb part for anyone who can't actually push more than 32 cores. But for those workloads that will scale well to 64 cores and 128 threads, it's going to be an awesome part.
Having competitive CPU cores while being a process node ahead gives you a lot of advantages. AMD has been able to exploit that with their third generation Ryzen and Threadripper parts, as well as their Rome EPYC server parts. Now they're going to be able to do the same with laptops.
For now, AMD is offering superior products at competitive prices, as they're trying to gain market share. They pretty much own the HEDT market, and have dominated the enthusiast portion of the desktop market for the last several months, though it takes much longer for server market share to adjust. But if they have years of high market share while Intel doesn't offer competitive products, AMD won't feel the need to offer such competitive prices forever. I'd like to see AMD make a lot of money this year, as I don't want them to have another near-death experience that risks leaving Intel and Nvidia as near-monopolies in their respective markets. But we really need Intel to get their act together and offer some better products within the next few years or else AMD is likely to become the dominant player in x86 and exploit that in many of the ways that Intel has. When multiple vendors offer competitive products, their customers are the real winners.
That analysis notably doesn't apply to Nvidia. While AMD is ahead of Nvidia on the process node of launched products, the only efficiency metric in which Nvidia is notably behind is die size--which not coincidentally, is the easiest thing to fix with a die shrink. Furthermore, Nvidia has access to (and commonly uses) the same process nodes as AMD, so there's no risk of Nvidia being stuck with a process node disadvantage over the long term. Intel does run that risk unless its foundry can get its act together soon.