Stories · as read on the AI station
The Wall of Silicon: Why Mining Hardware Slowed Down
Why a 7nm capacity crunch pushed miners to prioritize durability and heat management over yearly doubling, as InnoSilicon's CTO argued in 2019.
For six years, mining hardware followed a simple rule: every year, efficiency roughly doubled. At Mining Disrupt Miami 2019, held from July 23 to 25, 2019, the CTO of InnoSilicon told the audience that rule was about to break. When he spoke, Bitcoin traded at $9,770. On October 11, 2026, it was $83,093.
He started with his own company's history. InnoSilicon had been designing ASIC miners since 2013, when its first machine ran at about one terahash per second. By 2019, its latest Bitcoin miner used 10 nanometer technology from Samsung and reached 57 terahashes. That is about a doubling of hashrate every year, and he said the same was true across the industry.
Then came the catch. The state of the art was 7 nanometer, but as he put it, "the latest 7 nanometre technology is used heavily by cell phones and other 5 G's so capacity is the issue." InnoSilicon was still building mostly at 10 nanometer to avoid that crunch. His conclusion was direct: "we believe that efficiency improvement will slow down so it's not going to be doubling in hashrate or doubling in efficiency every year anymore." The company projected the pace of improvement would slow by at least 50 percent.
If machines are no longer replaced every year or two, they need to last. So InnoSilicon was repositioning its research toward durable miners that could run for two, three or four years. He walked through the T3: six fans in total, three hash boards, aluminum heat sinks throughout, and even a circuit board built on an aluminum substrate instead of the usual plastic, all to pull heat out of a box running a few kilowatts.
He then turned to the market. Crypto cycles, he said, last about four years, with a fast bull run and a long bear market. Hashrate cannot double overnight, because it takes physical machines and factory capacity, so it lags the price peak by about six months. That lag is where mining is most profitable, and the winners are the ones who buy machines at the start of a bull market. On the way down, hashrate falls almost in sync with price, because inefficient miners and miners with expensive power simply switch off. To survive a bear market, he said, you need an efficient miner and cheap electricity.
He closed by describing a partnership with HashShop, a new exchange where miners could sell their hashrate up front for terms of 90 to 360 days, taking some of the price risk out of mining.
The full story is on Intelligence Disrupt at intelligencedisrupt dot com, and many more like it play around the clock on Bitcoin Radio.