As the tensor cores are FLOATING POINT, while mining is an INTEGER function, as I mentioned back when the new Tesla was introduced I don't see them being a factor AT ALL for miners.

Interesting theory, but I'm still making enough to live on from mostly GPU mining.
Profitability IS down to where only folks with very cheap electric are making any money though.

With that said, I don't see the announced RTX 2070, 2080, or 2080 ti being good mining cards vs the 10xx series at their VERY HIGH PRICE, especially with the older cards starting to drop to pricing at or below where they were at 2 years ago before the big cryptocoin price runup happened - and the move of Bitmain, Innosilicon and others to put ASIC into most of what used to be the more popular GPU mined coins hasn't helped that any.

It will be interesting to see how these newer cards perform on Folding (which CAN be mined indirectly), but I suspect they're not going to be price/performance competitive even there.

We've seen this song and dance before - the whole point to a FPGA is that it can be reconfigured, so pointing at "the difference in architecture" is a meaningless statement.

I don't see them being ABLE to keep FPGA's at low performance for long it at all since a FPGA is MORE configurable than any GPU.

Won't have a significant effect, doesn't take any longer to rewrite the program for a FPGA than it does for a CPU or a GPU.
In fact, the only "drop" will be from miners that don't get updated in time for the fork I'm CERTAIN.

Scrypt has been UNprofitable to mine for any GPU since somewhere in 2014.

That's on raw parts though, need to see if the system-level tarriffs also changed - and gets complicated since most CPUs in particular are fabbed in the US or Germany (AMD via GF and Intel) and most miner chips are fabbed in Taiwan or Korea (TSMC and Samsung respectively) and should not be getting hit with these tarriffs at all.

I forget offhand on RAM but think most of that is fabbed in Taiwan or Korea or Thailand?

Tossup to a hair INFERIOR on efficiency at about the same hashrate as a well-tuned 6-card 1070/1070ti rig, perhaps a hair better efficiency than a well-tuned RX 470/480/570/580 based rig.
Lower cost but NO resale value long-term.

Did you ever get to play the multi-player "star trek" game the NTDS systems had?

24 TH for 2 KW?

Not very impressive if they're actually using 7nm parts.

Innosilicon has done better than THAT on new nodes....

I'm rather amazed they haven't made an offer for the shut-down Alcoa plant south of Wenatchee WA.

Or perhaps they made an offer and Alcoa hasn't agreed to it 'cause Alcoa has been talking about "perhaps" reopening that particular plant.

There are other equihash coins - might have been a shift to one of those.

Could also be some GPU miners pulling rigs due to low profitability, or shifting to other more profitable coins.

Could also be normal "luck" variation.

I suspect it's a combination of "all of the above" though.

If they're using one of the more recent upgrade process versions, it's possible.
They DO have a legit 28nm miner with efficiency that could argue with the Spondoolies SP20.

Also, NOBODY has 7nm or 10nm in production for a miner yet - eBang will probably be the first given announced plans, unless Bitmain or Inno switch to using Samsung for a fab.

The main reasons I won't even consider the Innosilicon T2 are:

 1) Minimum order size prices it WAY out of my reach.
 2) It's more efficient than the S9, but the cost is MUCH higher per hash - much more so than the small gain in efficiency.
 3) I can make more for the amount of $$$$ invested by getting other gear (that I CAN afford to get).

 4 through about 1000) I don't think it will make enough to pay for itself before the first of the 7/10nm generations miners hit the market and drive it into UNPROFITABILITY - even at MY low electric rate.

Only if you HAVE enough money to get close to that ceiling.

In some cases, money IS the limit.

Bitmain is selling quite a few D3 units (to use a specific example) to folks that CAN'T AFFORD 5x of the much more expensive (abet much better efficiency) Innosilicon unit.
A lot of miners don't HAVE $10,000 around to spend on miners in one purchace - and customs gets a LOT pickier about "large orders" like that as well.

 Can't "innovate" when your state-of-the-art product is already on the current "industry best" production node - which is WHY they have not released anything better than the S9 and probably won't for at least another year, no new node with significant performance upgrade has any capasity AVAILABLE to them, and it takes many months to a year to design a new chip on state-of-the-art nodes.

Innosilicon is on 14/16 nm where Bitmain is still at 28 nm on most of those "higher efficiency" pieces of gear, if not all.

It's telling that Innosilicon never bought the A3 to market (14/16 nm SHA256 miner that they HAD announced around the same time they announced the A4) - folks I know with connections have stated that it was because the A3 was NOT going to be comparable on efficiency to the S9, which is in line with everything else Innosilicon has released in the last 4 years or so.

It's also saying a lot that they only JUST RECENTLY finally managed to match the S9 with their new Terminator 2 unit - which is probably too little too late for them.

1500 at somewhere between 1000 and 1300 watts, on a 14/16 nm node process (possibly that "12nm" that's really an upgraded 16nm process).

900 wouldn't be worth it for them with the Innosilicon stuff around.

Like most websites, they have brief "overload" or "outage" situations.
May not be their fault - the Internet has core routers die every day, and it takes some time (sometimes seconds, sometimes minutes, rarely hours) for the dead router to get replaced/rebooted/worked around.

Already arrived, and the only one currently available turns out to be LESS efficient (by a hair) but somewhat lower cost than a well-tuned 6-card GPU rig.

Ignore those spam emails proporting to be from "bitmaintech", they are nothing but lies.