This though- this is very cool. If they've actually demonstrated 100 cycles in a pouch cell, made with reel-to-reel, that may mean that commercial cells are <5 years away. More likely 10+ years, as there are any number of things that could shut this down and the tech may not be compatible with cylindrical cells. Regardless, this is one of the most promising sounding press releases I've read in a long time. PVD anodes will be expensive (this is similar to the process to make mylar), but may be comparable to current processes, which require a long, temperamental and expensive solvent + drying process. PVD could even make this process more reliable (fewer failures) and consistent.
Commercial cells have 500-800 cycles in their lifetime (until their capacity falls to 80%) and are universally made on reel-to-reel machines. If it can't be done on reel-to-real it can't be done cheaply. There are a ton of difficulties moving from a coin cell to a prismatic/pouch cell to a cylindrical cell, but I can't understate how encouraging it is that they got to 100/400 cycles. It's near unheard-of with fully silicon anodes.
This is also quite promising for future development. When they say 1000-2000 mAh/g they're referring to the anode itself- only the weight of silicon, not the full battery. Silicon tops out near 4000 mAh/g, so there's a reasonable headroom there. 50% increase in overall capacity for the entire battery is fairly conservative. I presume it's because they can only apply very thin layers of anode silicon. That may mean there's a lot of room for growth though! They just have to thicken up that layer.
I'm still very skeptical of their long term capacity though. The problem with anodes like this is the nanoscale features. You basically have a huge tangle of velcro: that's done to increase the surface area exposed to the electrolyte, which solves the anode expansion problem. The drawback is that when that surface area becomes restricted, and it inevitably does, the SEI affects the distribution of li ions inside the silicon, increasing damage. They do appear to have found a way around that, but it may still put a long term limit on capacity. The thicker you try to make the anode (to increase energy density), the deeper the "velcro" becomes, and the more the SEI blocks lithium. It may also make these batteries more sensitive to heat and overcurrent and over/undervoltage- anything that disturbs the SEI may cause dramatic irreversible effects.
Random related fun fact: Silicon requires the use of copper rather than aluminum foils in batteries. Aluminum is a semiconductor dopant (the exact one used in your computer, in fact), and if you deposit silicon onto an aluminum foil it will form a very weak diode that causes a ton of problems. In computers an extremely thin silicon oxide film is used between aluminum wires and the transistors they connect. That layer is extremely resistive and causes a bunch of headaches, but way smaller ones than tiny diodes would!