CryptoSuper Report #16 — June 2026

This report is not, and must not be considered as financial, investment, or legal advice. The author holds no positions in Bitcoin mining stocks.

Executive Summary

For this 16th semiannual cryptocurrency supercomputing report, once again three coins make the cut for proof of work (hash power) generated annual economic value. In aggregate the annual minted value is around $12 billion for Bitcoin, Doge, and Zcash, and 93% of that is due to Bitcoin. No other POW coin exceeds $100 million in annual production.

Bitcoin had a significant pullback in value during Q4, falling by over 50% and apparently bottoming in February and revisiting that bottom this month. It seems to have resumed its long term climb and is trading around $66,000 currently. This pullback has squeezed miner margins and accelerated the pivot toward AI and HPC infrastructure hosting. They must switch to GPUs from ASICs in order to do so, but have the benefit of working at the efficient energy frontier and can offer competitive hosting rates.

Three nations: the US, China, and Russia dominate global hashrate with nearly 2/3 of all the hashrate between them, but there is also continued diffusion toward smaller countries with low-cost electricity inputs.

Hydrocooled systems have become increasingly important in the ASIC system mix, and individual machines are approaching the 1000 Terahash/sec or 1 Petahash/sec (10¹⁵) level in a global network that is of order 1 Zettahash (10²¹) sec. One sees immediately then that there are several millions of machines involved, considering a mix of older and newer systems.

Bitcoin mining is no longer merely “minting coins.” It is evolving into:

a planetary-scale energy balancing layer,flexible compute infrastructure,sovereign monetary infrastructure,and a thermodynamic monetization network.

from electricity and the cryptographic calculations of a continuous winner-take-all lottery, performed above ground.

Introduction

We produce the CryptoSuper 500 report twice a year, around the time of the two major supercomputing conferences: SC held in the US in November, and ISC held in May/June in Germany. This is the 16th report, representing 7.5 years of coverage. The prior 15th report from November 2025 is here:

https://cryptoassets0417.medium.com/cryptosuper-report-15-q4-2025-70fffc59b4a4

We track Bitcoin and a few other proof of work cryptocurrencies because they are a specialized form of high performance computing. Bitcoin is not AI, but it is a self-organizing autopoietic and persistent network for holding and transferring value that has endured for 17.5 years.

Why a Bitcoin Standard

Unlike other monetary tech Bitcoin defines a scientifically precise standard for supply that is also rooted directly in energy and math.

The algorithm has reduced supply emission every four years such that now each new block mints 3.125 Bitcoin. That reward will be cut in half again in two more years and the inflation rate will drop from 0.8% to 0.4%. There is an absolute cap on the supply of 21 million and there are less than 1 million BTC yet to be mined.

If Bitcoin were a company it would be the 13th largest by market cap; currently the network represents $1.3 trillion in value.

It is effectively the world’s largest decentralized supercomputer, consisting of millions of competing and cooperating systems, all running the same Nakamoto consensus algorithm in a continual lottery with a winner-take-all award for each block of approximately 10 minutes’ duration.

Gold comparison

Bitcoin’s price over its history has risen dramatically in gold terms, adhering to a power law steeper than the 5th power of Bitcoin’s age.

While gold holds its value through time, Bitcoin has outperformed gold and can also move very quickly in space, crossing the globe in an hour.

Since the conflict between the U.S. and Iran began on February 28, Bitcoin’s price has risen by over a third against gold; it has been a more effective geopolitical hedge during crisis. One reason is that Bitcoin is much more transportable, it can be sent anywhere across the globe within an hour (6 block confirmations). Iran has started accepting Bitcoin for tolls (fees) on ships passing through the Strait of Hormuz.

There is not yet strong public confirmation of large-scale crude oil purchases or settlements directly denominated in Bitcoin between sovereign states or major oil companies.

Figure 1. A log price vs. log age chart for Bitcoin measured in gold ounces. The solid line is a regression of a power law price (ounces) proportional to age to the 5.22 power.

Proof of Work coins

Out of 51 million (!) cryptocurrencies there are currently just 55 with over $1 billion market cap. Bitcoin’s market cap is $1500 billion, and there is an order of magnitude 1 Zettahash/sec of computational power devoted to decentralized Bitcoin minting, from several million Bitcoin systems around the globe. This is the world’s largest decentralized supercomputer.

Almost all cryptocurrencies and there issuers are lazy. They aren’t created via real work and energy consumption. They are air-dropped initially (created in thin air) and grow supply via proof of stake or similar algorithms. This is more akin to selling shares in a company and not at all like mining or minting.

Only 7 of the 55 are Proof of Work coins, using significant computational hash power to mint new coins and secure their networks and transactions. And of those, only three make our cut of over $200 million production per year, and are listed in Table 1. Failing to make the cut were Bitcoin Cash, Litecoin, Monero, and Ethereum Classic (an old variant that still uses hashing).

Table 1. POW coins with production of annual economic value exceeding $200 million. Bitcoin’s annual block production is $11 billion at $66,000 allowing for less than 1% fees and blocks slightly faster than 10 minutes. and if global hashrate is 0.95 Zettahash/sec that corresponds to $34 per Petahash/s per day gross revenue for a mining rig.

Total annual production is thus $11.8 billion including transaction fees which are running less than one percent of the block reward total. They are in the range of $0.086 billion per year.

Hashrate and Difficulty

Over the full history Bitcoin hashrate has grown as the 11th power of Bitcoin age but with significant variability. The growth was faster early on, amplified by the switch from CPUs to GPUs to SHA-256 customized ASICs followed by rapid advancement in the ASIC designs.

For the past number of years the rate of advancement appears to have slowed to around the 5th or 6th power of age, still very rapid.

Despite the slowing, the difficulty level is now around 125 trillion (dimensionless units) having started from difficulty equal to 1, and the present level corresponds to a global hashrate at the Zettascale level (10²¹ hashes/second). As of mid-June, 2026 the 30 day average is 0.95 Zettahash/sec.

Figure 2. The Difficulty as a function of Bitcoin block height. Blocks are spaced 10 minutes apart on average and the difficulty measure is proportional to the average hashrate over the prior 2016 blocks (about two weeks). Overall the hashrate has risen very steeply as the 11th power of Bitcoin block age, but has flattened in recent years as the mining industry matured.

Energy

Bitcoin miners have always pursued the lowest cost electricity inputs. The latest widely cited figures from the Cambridge Centre for Alternative Finance (CCAF) and its 2025 Cambridge Digital Mining Industry Report are:

Annual Bitcoin electricity consumption: ~138 TWh/year (average of 22 GigaWatts)Share of global electricity consumption: ~0.5%Sustainable energy share: 52.4% total sustainable sources: 42.6% renewables, 9.8% nuclearNatural gas share: 38.2% (now the largest single source)

Bitcoin mining is able to reach remote and stranded energy sources including natural gas and flared and vented methane, even landfill methane, and can mitigate atmospheric pollution. Hydroelectric sources are ideal, along with wind and solar in remote locations. Bitcoin mining rigs can be dropped in adjacent to the sources since low-bandwidth Internet connections are sufficient.

Bitcoin mining was once severely criticized for energy usage, but it has never reached 1% of global electricity and the concerns have shifted to data center requirements. Bitcoin inputs are greener than datacenter at this point.

Daniel Batten is one of the best voices around promoting Bitcoin’s role in stabilizing energy grids. He argues — quite successfully — that Bitcoin mining should be viewed not merely as electricity consumption, but as a flexible, interruptible energy infrastructure layer that can stabilize grids, monetize stranded energy, absorb renewable overbuild, and improve the economics of new power generation projects.

“Because Bitcoin mining is the only energy user in the world where a gargantuan 80% of operating costs are spent on power, Bitcoin mining units are generally the only business for whom it is profitable to seek and destroy methane and use it for power in remote places with poor Internet connectivity, where the capital costs to utilize that methane are high.” — Daniel Batten

Mining is increasingly viewed not merely as:

energy consumption,

but also as:

flexible interruptible load,synthetic demand response,and economic storage for stranded or intermittent energy.

The literature and utility discussions now frequently frame mining as:

“virtual battery” infrastructure,especially for renewables.

This is a major conceptual shift compared to the older environmental critique era.

From the Hashrate Index of 5/21/26: “Texas hosts a significant share of global Bitcoin mining hashrate — approximately 17% as of mid-2026. When a large fraction of that capacity curtails simultaneously to avoid a 4CP peak, the effect shows up across multiple on-chain signals…ERCOT’s Four Coincident Peak (4CP) mechanism sets a site’s transmission charge for the next year based on its average load during the four highest 15-minute system peaks across June–September.”

Bitcoin miners are able to rein in their energy usage much more quickly than other loads when there is stress on the grid, their load is highly modular, and it is self-funding.

Mining Equipment

China continues to dominate in the design of ASICs for Bitcoin minting systems, from companies including Bitmain, Canaan, and MicroBT. The fabs though are located in Taiwan (TSMC #1) and Korea (Samsung #2).

Table 2. From ASICminervalue.com, latest models from three companies, in the range of 300 Terahash/sec to 1160 Terahash/sec. The first Petahash/sec systems are appearing. Almost all models in the table are liquid cooled. Power requirements are substantial, ranging from 3500 to 11000 Watts.

The leading edge for Bitcoin minting ASIC systems is now heavily hydro/liquid cooled:

Bitmain Antminer S23 and S21 Hydro,Bitdeer SealMiner series,and similar platforms.

The optimization target has shifted from:

raw TH/s
toward:Joules per terahash,thermal density,rack-scale cooling efficiency,and power orchestration.

Mining increasingly resembles HPC engineering, which is one reason why Bitcoin mining farms, along with their low electricity costs, are well-suited to host AI workloads as well.

Figure 3. AI Data Center with Integrated Bitcoin Mining as Flexible Load. Conceptual architecture showing how a modern AI data center can allocate power between mission-critical AI workloads and a dedicated Bitcoin mining wing that functions as interruptible, easily curtailed demand. Core AI training, inference, and high-performance computing workloads receive priority access to electricity, while Bitcoin mining consumes excess capacity and can rapidly reduce consumption during periods of grid stress or elevated demand. This arrangement enables higher overall infrastructure utilization, monetizes stranded or surplus energy, supports renewable generation integration, and provides grid-balancing services. The result is a symbiotic relationship in which AI drives demand for data center capacity, Bitcoin mining provides operational flexibility, and the electric grid benefits from a large-scale controllable load resource.

Mining Distribution and Companies

According to hashrateindex.com the US has 37% of global hashrate (1/3 or more of that 37% in West Texas), Russia 16%, and China 12%. Collectively the top three nations are responsible for nearly 2/3 of all hashrate. Other significant countries in the top 8 are Paraguay with 4%, Ethiopia, the UAE, Oman, and Canada with 3% each. That already adds up to 80% of the hashrate being found in 8 countries.

But this is a substantial improvement in decentralization, as in the past China had been responsible for over 50% of global hashrate; they cracked down 5 years ago due to energy concerns and capital flight concerns, and perhaps to help promote their own digital currency. Hashrate following those restrictions moved especially to the US and Russia.

Over 25% of the hashrate is operated by publicly traded companies, mostly in the US. The top 10 are tabulated in Table 3 and collectively have 279 Terahash/sec. (not all of which is operating simultaneously).

Table 3. Top 10 publicly traded Bitcoin mining companies. Shown are market cap, total current hashrate available and recent news items relating to activity in AI and HPC. The companies’ collective market cap exceeds $40 billion and they are involved in a number of new deals amounting to well over $20 billion of fund raising.

There is a large pivot underway for these companies to add AI and HPC capacity to balance out their revenue streams and earn higher margins. A few structural observations for these companies stand out:

1. The industry is increasingly bifurcating into pure Bitcoin miners, and hybrid “power + compute infrastructure” operators.

2. AI/HPC revenue is becoming material very quickly. CoinShares projections cited in 2026 reports estimate that firms like Core Scientific, IREN, and TeraWulf could derive a majority of revenue from AI/HPC workloads within a year or two.

3. The key scarce asset is no longer merely ASIC fleets; it is:

grid access,substations,cooling infrastructure,and dispatchable megawatt-scale power.

Bitcoin miners already possessed those assets. The AI pivot monetizes them at much higher revenue multiples than SHA-256 mining alone.

4. Hashrate leadership currently remains concentrated in:

MARA,CleanSpark,Riot,IREN,and Core Scientific.

5. The AI pivot is beginning to reduce marginal Bitcoin-network expansion pressure because some energy capacity is being diverted from SHA-256 mining into GPU compute.

The largest deals recently:

Core Scientific transformed itself from a distressed Bitcoin miner into a premier AI/HPC infrastructure platform through massive long-term hosting agreements and its strategic relationship with CoreWeave. Deal size: $10 billion.IREN Limited executed one of the sector’s largest AI pivots by leveraging its renewable-powered mining campuses into hyperscale AI cloud infrastructure tied to partnerships involving Microsoft and NVIDIA. Deal size: $10 billion.TeraWulf repositioned its energy-intensive mining operations toward AI compute and data-center leasing through major HPC agreements and partnerships with Fluidstack and Core42. Deal size: $4 billion.Hashrate Index on 5/22/2026 wrote “Miners with existing data center infrastructure, power relationships, and cooling capacity are positioned to participate in the AI infrastructure buildout as this market validates. Whether through mullet mining hybrid operations, direct GPU and ASIC colocation for neoclouds and enterprise AI customers, or full transition to AI compute infrastructure, the optionality is more valuable than it appears at first glance. The AI chip sector’s $100 billion-plus annual CapEx has to land somewhere physical. Bitcoin mining operators own a lot of the physical infrastructure it needs.”

The Bitcoin balances held by miners have been decreasing of late as they pivot toward AI. They have sold some of their Bitcoin as needed for ‘skin in the game’ as part of new financing packages for AI infrastructure buildout.

Table 4. Bitcoin balances overall held by Bitcoin mining companies have fallen during the past 6 months (tabulated by bitcoinminingstock.io). Most notably, the two largest holders among these have sold significant portions of their balances as part of their pivots toward AI. These balances are modest relative to those held by ETFs, treasury companies, and several governments.

BTC in Treasuries

The net buying of Bitcoin has been primarily through ETFs and Treasury companies in recent years, mostly as custodians for their clients and shareholders. The amount held in these companies and funds has been rising faster than the fourth power of Bitcoin’s age.

Currently ETFs and exchanges hold over 1.6 million Bitcoin, some 8% of the 20 million Bitcoin minted to date.

Public companies hold 1.26 million Bitcoin according to bitcointreasuries.net

Strategy alone holds over 840,000 Bitcoin, which is 4% of the maximum 21 million supply.

Governments hold 650,000 or 3% of supply, with the top three being the US with over 300,000 BTC, China with 190,000 BTC, while the UK and Ukraine hold about 61,000 and 46,000 respectively.

Private companies including BlockOne, Tether, and SpaceX hold around 282,000 BTC and some 380,000 is locked into DeFi protocols.

Still the vast bulk of Bitcoin appears to be in private hands and individuals are the beneficial owners of ETF holdings and treasury company holdings for the most part.

Supercomputing Comparison

In the 15th report I compared the Bitcoin decentralized supercomputing minting network to the world’s fastest computer, El Capitan, and to the Top 500 list of supercomputers.

El Capitan has a peak performance of 1.74 Exaflops (the unit is a billion billion floating point operations), and consists of 11 million GPU and CPU cores. It consumes 30–35 MegaWatts, which is of order 10,000 times as much as a typical Bitcoin mining rig. The cost is $600 million and the science mission is of critical national importance.

But it would be of little use as a crypto mining system. How fast could it hash Dogecoin? The Scrypt algorithm is designed to be highly memory intensive, and the Scrypt ASICS have their entire die as essentially one giant SRAM scratchpad with a tiny compute wrapper around it. It is estimated that El Capitan would mine roughly $47 worth of Dogecoin per day — less than a single Antminer L9 ASIC (~17 TH/s, ~$14,000) optimized for Scrypt. The memory hierarchy mismatch is absolutely brutal for Scrypt on HPC silicon.

How about Bitcoin’s doubled SHA-256 algorithm? It is about 75% bitwise logic and 25% 32-bit integer. Adopting generously 3 Exaops peak, and noting there are 1600 ops approximately in the the double SHA-256 hashing round, we are looking at less than 2 Petahashes/sec. This is comparable to only 2 to 4 ASIC mining rigs from Table 2. More than a million such of the very latest ASIC hashing systems would be required to produce the global hash rate.

Bitcoin ASICs are multi-module, highly pipelined, and very highly parallel in replicating functional units, and also gain enormous advantage because they:

hardwire the exact dataflow,eliminate instruction decode,eliminate caches,eliminate memory hierarchy overhead,eliminate branch/control logic,deeply pipeline the exact rotate/XOR/add structure,and maximize hashes/Joule rather than generic ops/sec.

There is a profound security statement here — the economic and thermodynamic investment embodied in the Bitcoin network’s hashrate is now so large that even nation-state supercomputing infrastructure is irrelevant to it. The network’s security is anchored to cumulative energy and highly specific ASIC hashrate expenditure, not to any single machine’s compute ceiling.

It’s also worth noting that Bitcoin miners crossed the Zettahash threshold (~10²¹ hashes/sec) quietly and with little fanfare, while the HPC community is still celebrating Exascale (~10¹⁸ FLOPs) as a frontier achievement. Bitcoin’s proof-of-work has outpaced civilian supercomputing by three orders of magnitude — a remarkable emergent property of the self-organizing nature of the Nakamoto consensus.

Quantum Risk

There are two areas of risk from quantum computing for the Bitcoin network. First one must remark that quantum computing is a risk to the entire financial system and a number of quantum safe key mechanisms already exist, however they are more expensive to implement.

One risk is many decades away, and that would be Grover’s algorithm allowing quantum computers to generate SHA-256 hashes. But the clock rates of today’s quantum computers are very slow so this is not a realistic threat for a long time to come. And at the end of the day it would just be another set of systems mining Bitcoin and processing transactions on the network, not a way to steal coins.

The more immediate risk in the 5 to 20 year timeframe (and it is that uncertain) is from Shor’s algorithm. It has the potential to threaten 256-bit keys once there is a machine with of order 1000 logical qubits able to crack a public key during the 10 minutes that transactions are sitting in the mempool waiting for processing.

Shor’s algorithm would derive a private key from an exposed public key. Early Bitcoin output transactions from before 2012 expose the public keys at all times (P2PK) but since then most transactions only expose keys while they are being spent, within a single block duration. An important practice is not to reuse addresses; instead generate a new address for each transfer.

Glassnode Research estimates 9.6% of issued supply is exposed through script structures for which public keys are visible on the blockchain.

Current estimates are that it will take 500,000 to 1,000,000 or more physical qubits in order to have enough coherently entangled logical qubits for Shor’s algorithm to succeed. There has been very active progress in reducing the physical to logical ratio, and that needs to be monitored.

But there are already multiple NIST approved quantum-safe key algorithms, it is primarily a matter of migrating Bitcoin and keys to quantum safe wallets. A number of BIPs (Bitcoin improvement proposals) are in various stages of discussion and progress to address the issue.

Very old coins that have not moved, such as Satoshi’s 1 million coins spread across a large number of wallets, remain vulnerable and there is debate as to whether to freeze or throttle those or just let them sit as an early warning system, the potential trillion dollar canary in the coal mine if Bitcoin’s price keeps following its long term steep power law trajectory.

Outlook

The Bitcoin mining industry is entering a new phase of its maturation. The large publicly traded miners are increasingly evolving into hybrid energy, AI, and high performance computing infrastructure companies rather than remaining pure-play SHA-256 minting businesses. Ironically, this pivot may create more opportunity for decentralization within Bitcoin mining itself. As large firms divert portions of their capital expenditure and electrical capacity toward GPU clusters and AI hosting, smaller and mid-sized operators may regain competitiveness in SHA-256 mining. At the same time, nation-states appear increasingly likely to expand their participation, whether through sovereign reserve accumulation, domestic mining initiatives, grid-balancing applications, or strategic energy monetization.

Bitcoin’s volatility has clearly been declining, roughly in inverse proportion to Bitcoin’s age, despite periodic severe drawdowns. This moderation is important because it alleviates pressure on miners, treasury companies, lenders, and custodial institutions. A less volatile Bitcoin is easier to finance, easier to collateralize, and more compatible with sovereign and institutional balance sheets. If Bitcoin continues transitioning from a speculative emerging asset toward a global reserve settlement layer, declining volatility may ultimately prove more important than short-term price appreciation alone. Indeed, declining volatility allows for larger and more risk conscious tiers of capital to participate.

The next halving, expected in Q2 2028, will reduce the block subsidy from 3.125 BTC to 1.5625 BTC per block and lower Bitcoin monetary inflation from approximately 0.8% now down to 0.4% annually. Historically each halving has intensified competitive pressure within mining while simultaneously reinforcing Bitcoin’s scarcity narrative. The coming cycle may be particularly interesting because it will occur in a world where ETFs, treasury companies, sovereign entities, AI infrastructure operators, and energy markets are now deeply entangled with the Bitcoin ecosystem.

Quantum computing remains a long-term engineering concern rather than an immediate existential threat. Current evidence suggests that practical attacks against exposed Bitcoin public keys are unlikely to become operationally significant until sometime in the next decade, and even then only for a subset of vulnerable legacy wallets and address structures. Multiple quantum-safe signature systems already exist, and the Bitcoin ecosystem has substantial time to migrate toward more secure cryptographic standards. In practice, quantum risk is not unique to Bitcoin; it applies broadly to the global financial, governmental, and Internet security infrastructure.

Bitcoin minting will continue to become more integrated into the global energy infrastructure.In Daniel Batten’s view: “The future of Bitcoin mining is that it seeks out stranded energy at near zero marginal cost. The future of Bitcoin mining is that it stabilizes grids and recycles heat as a primary revenue source, with the block reward and fees being the by-product. The future of Bitcoin mining is that it solves hard problems the world desperately needs — keeping our grid safe and stable, making AI load flexible, reducing harmful methane pollution, stopping the wasteful practice of renewable energy curtailment in ways that benefit whole communities.”

More broadly, Bitcoin mining is beginning to resemble a new category of industrial infrastructure altogether: part monetary network, part energy-balancing layer, part distributed supercomputing fabric, and part geopolitical reserve system. The network’s scale now exceeds civilian supercomputing by several orders of magnitude in raw cryptographic throughput, while simultaneously integrating itself into electrical grids, capital markets, sovereign strategies, and AI datacenter expansion. What began as an experimental peer-to-peer currency has evolved into a planetary-scale thermodynamic and computational system whose long-term implications are still only beginning to emerge.

---

CryptoSuper Report #16 — June 2026 was originally published in The Capital on Medium, where people are continuing the conversation by highlighting and responding to this story.