so lets imagine a scenario:

network was 200exahash

every 16 blocks (6.25btc*16=100btc)
200 exahash shared 100btc every 2hours 40 minutes
but lets say it costs bitmain asics $27k a coin to mine previously
where by 100coins later (2hours 40 minutes) they spent $2.7m
.. or for those that like detail and a bit more accuracy
(pencil math: 10 exahash is 5btc, 100peta is 0.05btc, 1peta is 0.0005btc, 100terra is 0.00005)

and so a 140terra bitmain asic is 0.00007btc($2.10 at $30k/btc) every 2 hours 40min income

now lets work out the outgoing cost
the asic costs $11.6k. which is spread over 2 years
(pencil math: $11.6k/730, $15.89/day, $0.6621 an hour, $0.2207 every 20minutes)
meaning hardware cost is: $1.7656 every 2 hours 40 minutes
imagine electric cost is 0.04/KWH which at 3kw/h is $0.3192 for 2 hours 40minuts
meaning bitmains costs are $2.0848

thus bitmain asics are only just breaking even at $0.04/kwh

imagine intel jumped in.. and added 10exahash.

bitmain are still spending $2.08 per hour..
but now with 210exahash
means 10exahash gets a share of the 100btc earned per 2 hours 40mins

meaning the 200exa of bitmain are not getting 100btc. they are only getting 95btc.
which if you share it down.
means each bitmain asic is not getting 0.00007btc but instead 0.0000665($1.99 at $30k/btc)
so now bitmain asics are losing $0.0948 every 2 hours 40 minutes

.......
but here is the rub...
before intel jumped in. bitmain asics may have sold even at break even 100btc every 2 hours 40minutes to the markets..
they are now at a loss. meaning they are not going to sell their 95coin share to the markets. thus the push down on markets is reduced by 95btc every few hours less sells occuring.

bitmain now have an average cost of $31,350/btc so wont sell for less than this.
if the entire open mining network cannot make coin anymore for $30k and instead it costs them $31,350 for the most (previously efficient) miners to mine. then everyone has a decision to make.

a few asic miners turn off their asics. and instead buy coin at $30k rate instead of spending value at $31.35k rate by mining. (cheaper to buy than to mine)
where there is more market buying with less market selling

which then causes a decrease in hashrate as there are less asics mining.
and an increase of market price

which can cause the hashrate competition for the 100btc back into alignment of the shares of the reward covering the costs because if less people are taking from the reward, more reward goes to each remaining miners.
..so in this instance its not a race to the top or bottom, its rebalancing