Bitcoin is going through its first-ever “hashrate bear market,” according to Twenty One Capital CEO Raphael Zagury — the longest stretch in the network’s history where total mining computing power has failed to recover its previous record high.
By Marcus Johnson | September 2, 2026
Zagury made the argument at the Bitcoin Asia conference in Hong Kong on August 28, and the Tether-backed Bitcoin treasury company subsequently filed the prepared transcript with the US Securities and Exchange Commission. His presentation materials, cited by crypto.news, calculate that Bitcoin’s hashrate has fallen roughly 22 to 24 percent from its late 2025 peak. For everyday investors, this is a story about who secures the Bitcoin network — and why the economics of that job are changing fast.
The Hook: What Hashrate Is and Why It Matters
Hashrate measures the estimated computing power that miners contribute to securing Bitcoin and competing for block rewards — think of it as how many workers are on shift at the global mine. More power generally means more machines running and a more secure network. Bitcoin’s hashrate approached 1.3 zettahashes per second late last year before entering a prolonged decline, Zagury said.
Worth noting: “hashrate bear market” is Zagury’s description of the current cycle, not an official network classification. Bitcoin does not publish an exact count of active machines, so analysts infer hashrate from block production rates and mining difficulty — which means daily readings can swing sharply. CoinWarz estimated the hashrate at about 829 exahashes per second on September 2, after readings moved above one zettahash on several days in late August.
On-Chain Evidence: The Numbers Behind the Slump
- 22 to 24 percent — estimated hashrate drawdown from the late 2025 peak, per Zagury’s presentation
- 829 exahashes per second — CoinWarz hashrate estimate for September 2
- 19.9 percent — decline in Bitcoin mining difficulty from its November peak by late July, per prior analysis
- 287 days — approximate length of the hashrate downtrend as of late July, per Bitcoin Magazine Pro data
When fewer machines compete, the network automatically adjusts mining difficulty downward — and that is the quiet silver lining for remaining miners. Lower difficulty means each surviving machine earns a larger share of the block rewards. Previous analysis cited by crypto.news found mining difficulty had already fallen 19.9 percent from its November peak by late July, with hashrate in a downtrend for roughly 287 days at that point.
The Core Conflict: Bitcoin Mining Now Competes With AI
Zagury contrasted the current decline with the shock of China’s 2021 mining ban. Back then, hashrate collapsed rapidly as companies shut down Chinese facilities — but it recovered once the same machines relocated to North America, Central Asia and elsewhere. The machines simply moved.
This cycle is different, and more gradual. Rather than relocating equipment, operators are reconsidering whether new electricity and data center capacity should go to Bitcoin mining at all. The reason is artificial intelligence. “This has been the longest period that we’ve seen from an all-time high until recovery,” Zagury said, adding that AI creates a competing use for miners’ power capacity and infrastructure worldwide.
“If you look at the public mining companies out there, there really isn’t anybody staying the course to mine Bitcoin at scale,” he said. “Pretty much everybody is leaving the industry right now.” That claim describes a broad trend rather than a literal exodus — MARA, CleanSpark, Riot, Bitdeer and other publicly traded companies still operate large Bitcoin mining fleets even as some explore or build AI infrastructure. The pivot is most advanced at companies such as TeraWulf, IREN, Core Scientific, HIVE and Cipher; TeraWulf has reported 21 million USD in AI and high-performance computing hosting revenue.
The two industries compete for the same scarce resources: large power connections, cooling systems, land, data center buildings and access to capital. Converting a mining site to AI is not as simple as swapping chips — AI facilities need different networking equipment and construction standards — but sites with secured power and fiber access offer a head start for high-performance computing.
Market Implications: What It Means for Your Bitcoin
Does a shrinking hashrate put your Bitcoin at risk? Not in any immediate sense. Bitcoin remains secure at current levels, and a 22 to 24 percent drawdown leaves the network far larger than it was just a couple of years ago. The deeper question is long-term: Bitcoin’s security budget depends on miners finding the business worthwhile, and AI is now bidding away their electricity.
For investors, there are two practical takeaways. First, miners that survive the shakeout get a larger slice of rewards as difficulty falls — a quiet boost to the strongest hands. Second, the AI pivot explains why many mining stocks no longer track Bitcoin’s price the way they used to; their revenue increasingly comes from hosting computing power, not mining coins.
The Verdict
Zagury’s “hashrate bear market” label is a framing, but the data behind it is real: record highs set in late 2025 remain unreclaimed, difficulty is down sharply, and capital that once expanded mining fleets is being diverted to AI data centers. Bitcoin the asset is holding firm around 77,000 USD while Bitcoin the industry undergoes its most significant restructuring since the China ban. How — and whether — hashrate recovers will be one of the defining stories of this cycle.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
22% down and difficulty adjusts with it, smaller miners get breathing room. the article gets that part right
breathing room until the ai pivot guys come back for the next difficulty jump. this cycle just repeats
breathing room until the next diff jump squeezes them right back out. watched the same thing in 22, only the hydro operators were still standing by winter
same as 22 exactly. by spring the efficient fleet owned more of the network than before the drop, difficulty gifts rebuild fast
agreed, and Zagury basically says so himself. the transcript went straight into the SEC filing, bears keep skipping that part
we used to measure hashrate in terahashes on a usb stick. either way security is still way beyond any realistic attack budget
22% off peak and people act like the network is dying. Hashrate chasing AI revenue is just miners being rational, security is still fine at these levels.
Rational until the AI datacenter contracts dry up and they crawl back to sha-256. This exact pivot happened before.
@Minjun P. the AI contracts drying up is the flip side nobody prices. half those michigan style conversions will crawl back to sha256 at the next difficulty gift
they crawl back at whatever difficulty the AI datacenters left behind too. happens every pivot, the sha256 door never closes
longest stretch ever from ATH to recovery and price is still holding 77k. the miners leaving were the marginal ones from last cycle anyway
marginal ones yeah, s19 fleets running on cheap flare power were first out the door. the difficulty drop basically gave the survivors a raise
A 22 percent hashrate drop landing while price holds 77k is actually the healthy version. The miners leaving are the ones who bought rigs at the top with debt.
and those debt-funded buyers are exactly who exits first in a 22 percent reset. Zagury filing the deck with the SEC means the data is public, not just conference chatter
Security still beyond any realistic attack budget, and yet the news cycle calls it a bear market. Difficulty adjusts, it always does.
still beyond any realistic attack budget even after losing a quarter of it. 1.3 zettahash was itself an absurd number two years ago
sure, but the attack budget math ignores who pays for security. thinner hashrate means thinner fee split for whoever stays, and the difficulty gift only lasts until the next retarget
the attack budget point holds but the fee split concern is the real one. security spend gets subsidized by block subsidies that halve again in 2028, thinning hashrate right into that is the actual experiment
s19 fleets on flare gas exiting first makes sense, their margins were cents per kWh anyway. the 22% reset is a survival filter, not a crisis
flare gas s19s were marginal even at 1.3 ZH. calling it a survival filter is right, the network is pruning not bleeding
the pruning framing is the right one. the fleet that survives a 22 percent reset is the one you want securing blocks anyway
1.3 zettahash down 22 percent and security spend still dwarfs every other chain combined. the headline sounds scary, the math does not
from 1.3 zettahash down 22 percent and security spend is still absurd. the framing of first hashrate bear market is fair but the panic is overdone
Zagury calling it a bear market while his own treasury thesis depends on security holding. At least the SEC filing forces the data into the open, 22% from peak is survivable.
the detail nobody mentions: he said this at Bitcoin Asia on Aug 28 and the deck was filed with the SEC days later. Twenty One Capital wants the hashrate narrative framed before their own Nasdaq run