For years, the story of mining hardware was simple: every new generation roughly doubled the hashrate of the last. At Mining Disrupt Miami, recorded July 23–25, 2019, the CTO of InnoSilicon took the stage to argue that this era was ending. Efficiency, he said, would no longer double every year, and the priority was shifting toward durability.
The 7nm Bottleneck: Why Phone Chips Are Hiding from Miners
He began by tracing InnoSilicon's ASIC designs, noting that the company had been designing miners since 2013, when it released the A1 at about one terahash per second. In 2019, its latest generation of Bitcoin miners used 10 nanometer technology from Samsung to reach 57 terahashes per second. Over those six years, he said, miner efficiency across the industry had been doubling about every year as chips moved from 20 nanometer to 16 and then down to the 10, 8 and 7 nanometer region.
However, he introduced a critical caveat. The 7 nanometer node was the state of the art, but its capacity was largely taken by cell phones and 5G. As a result, InnoSilicon was still manufacturing mostly at 10 nanometer to avoid the capacity issue at 7.
the latest 7 nanometre technology is used heavily by cell phones and other 5 G's so capacity is the issue
InnoSilicon's CTO · 3:49
Because miners were already at the state of the art, he said, efficiency gains would slow. InnoSilicon was projecting that efficiency improvement would slow by at least 50 percent from that point on.
From Speed to Survival: Designing Machines for Four-Year Cycles
With the promise of annual doubling receding, the design philosophy for ASICs had to change. Instead of replacing machines every year or two, the goal was a miner that could run reliably for two, three or four years.
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
InnoSilicon's CTO · 4:22
He said InnoSilicon was repositioning its research and development effort to build more durable miners.
we are not looking at the replacing - every year or every two years we want the manor to be more reliable
InnoSilicon's CTO · 4:38
The Heat Problem: Why Aluminum Beats Plastic in ASIC Design
The physical side of this shift was visible in the T3 miner, which he walked through on stage. The power supply has two small fans and the main unit has four, two in front and two in back, for a total of six, all there to improve airflow and carry heat away as quickly as possible. Inside are three hash boards in a vertical arrangement, and the machine is almost all metal, with a lot of aluminum heat sinks.
He pointed out that most PCB boards use FR4, a plastic, but InnoSilicon chose an aluminum substrate for the board to help with heat dissipation. The board attaches to an aluminum sheet, and aluminum heat sinks sit on top of the chips as well. The key in miner design, he said, is thinking about how to remove the heat.
you have a few kilowatts inside that little container that you have to manage the heat
InnoSilicon's CTO · 8:19
Durability and heat management mattered because of the market cycle that every miner has to survive, which was the next part of the talk.
The 6-Month Lag: Why Early Buyers Win the Bull Market
Moving from hardware to market dynamics, he described a crypto cycle of about four years, with a fast bull run followed by a long bear market. In the 2017 and 2018 cycle, he noted, the price peaked around 20k US dollars and was followed by about two years of downtrend.
Hashrate behaves differently from price. Because hashrate requires physical machines and manufacturing capacity has to ramp up, it cannot double overnight. Even with the latest technology, he estimated, doubling the network hashrate would take one to two million new machines. That is why, he said, there is typically about a six month lag between the price peak and the hashrate peak, and why pre-order times for new miners were already two to three months.
the most profitable time to do mining is the between this time lag
InnoSilicon's CTO · 10:55
The key, he said, is to start mining quickly at the start of a bull market. On the way down, it works the other way: when the price fell from about 6,000 dollars to about 3,000 dollars around the end of 2018, hashrate dropped almost in sync, because miners with less efficient machines or expensive electricity simply shut off. To survive a bear market, he concluded, you need an efficient miner and cheap electricity.
Hashrate as a Commodity: Selling Future Output on Hashshop
Near the end of the talk, he described a partnership with HashShop, a hashrate exchange that had come online about two weeks earlier. The platform let miners monetize their hashrate up front by selling it for terms from 90 to 360 days, instead of mining for a year and carrying the Bitcoin price risk.
He said the platform allowed both buyers and sellers and focused on long term contracts, unlike other exchanges centered on instantaneous hashrate. The goal was to remove much of the uncertainty of mining and Bitcoin price swings.
He also mentioned that InnoSilicon designed miners for other coins and that the company mined itself, with facilities in China, Kurdistan, Canada and the U.S. The conversation on mining hardware continues at Intelligence Disrupt 2027, March 22 to 24 in Irving, Texas. When this session was recorded, Bitcoin traded at $9,770; on October 11, 2026 it was $83,093.