Bitcoin Is in Its First Ever Hashrate Bear Market, Twenty One Capital CEO Says

Bitcoin Is in Its First Ever Hashrate Bear Market, Twenty One Capital CEO Says

Bitcoin is experiencing its first ever “hashrate bear market,” according to Raphael Zagury, chief executive of Tether-backed Bitcoin treasury company Twenty One Capital, who argued at Bitcoin Asia in Hong Kong that the network’s computing power has failed to reclaim its late 2025 record because miners are redirecting capital toward artificial intelligence instead.

Zagury, who presented the thesis on Aug. 28 and subsequently filed the prepared transcript with the U.S. Securities and Exchange Commission, said Bitcoin hashrate approached 1.3 zettahashes per second late last year before entering a prolonged decline. His presentation materials calculate a drawdown of approximately 22 to 24 percent from that peak.

The phrase “hashrate bear market” is Zagury’s description of the current cycle rather than an official network classification. It refers to the unusually long stretch during which estimated computing power has failed to return to its previous all-time high. “This has been the longest period that we’ve seen from an all-time high until recovery,” he said.

A different kind of decline

Bitcoin hashrate measures the estimated computing power that miners contribute to securing the network and competing for block rewards. The metric matters because it reflects both network security and the level of industrial investment flowing into Bitcoin mining.

Zagury explicitly contrasted the current cycle with the shock produced by China’s 2021 mining ban. In that episode, hashrate collapsed rapidly as companies shut down Chinese facilities, but it recovered once the same machines were relocated to North America, Central Asia and other regions. The machines, and the industry built around them, survived; only their geography changed.

This time, the decline has developed gradually, and the underlying dynamics are different. Rather than relocating existing hardware, operators are reconsidering whether new electricity capacity, land and data center buildouts should be allocated to Bitcoin mining at all. AI has created a competing use for the exact resources miners control: large power interconnections, cooling systems, data center buildings and access to capital.

“If you look at the public mining companies out there, there really isn’t anybody staying the course to mine Bitcoin at scale,” Zagury said. “Pretty much everybody is leaving the industry right now.”

That statement describes a broad trend rather than a literal exodus. MARA, CleanSpark, Riot, Bitdeer and other publicly traded companies continue to operate large Bitcoin mining fleets even as they explore AI infrastructure. But the pivot is most advanced at companies such as TeraWulf, IREN, Core Scientific, HIVE and Cipher. TeraWulf reported 21 million USD in AI and high performance computing hosting revenue in the first quarter, exceeding its Bitcoin mining revenue for the first time. Cipher has secured a 200 million USD revolving credit facility to finance its expansion into long-term AI data center contracts.

The numbers behind the claim

Network estimates vary because Bitcoin publishes no exact count of active machines; analysts infer hashrate from block production rates and mining difficulty, and daily readings fluctuate sharply. CoinWarz estimated hashrate at roughly 829 exahashes per second on Sept. 2, after readings moved above one zettahash on several days in late August. Longer moving averages, which smooth out statistical noise, provide the clearer picture, and on that view the downtrend that began after the late 2025 peak remains intact.

The difficulty data corroborates the story. Bitcoin adjusts mining difficulty every 2,016 blocks, roughly every two weeks, to keep average block production near ten minutes. Previous analysis found difficulty had fallen 19.9 percent from its November peak by late July, with hashrate in a downtrend for approximately 287 days according to Bitcoin Magazine Pro data.

Why staying power can pay off

Zagury rejected the idea that Bitcoin mining is inherently a bad business. Profitability, he argued, depends on where an operator sits on the industry cost curve. A miner running efficient equipment on cheap electricity can remain profitable in conditions that force higher-cost competitors to shut down, while heavy debt loads and short repayment schedules can sink even operationally sound facilities.

Hash price, the measure of expected miner revenue per unit of computing power, remains low by historical standards, which puts operators running older machines or expensive power contracts under sustained pressure.

Yet a shrinking network carries a silver lining for those who remain. When computing power leaves the network, Bitcoin’s difficulty adjustment mechanism eventually responds downward, making it easier for surviving miners to find blocks. Each operator that stays active controls a larger share of network hashrate and block rewards without buying a single new machine.

“The beautiful thing about Bitcoin mining being in a bear market of hashrate is that, for those that stay around, they naturally get a higher share of the market,” Zagury said.

That benefit is not a guarantee of profit. Revenue still depends on Bitcoin’s price, transaction fees, electricity costs and fleet efficiency. But the mechanism explains why some operators may see the current cycle less as an existential crisis and more as a consolidation opportunity.

A structural shift, not a repeat of 2021

The deeper implication of Zagury’s framing is that this cycle may not mean-revert the way previous hashrate shocks did. As long as AI workloads offer higher returns on power and data center capacity than Bitcoin mining, capital that would once have expanded the network’s computing base will flow elsewhere. For an industry that has spent a decade assuming hashrate only goes up, that is a genuine change in regime, and one that investors in both mining equities and Bitcoin itself will be watching closely in the quarters ahead.