Forecasting Bitcoin Hashrate in 2030
- Jason Deane
- 3 days ago
- 8 min read
Updated: 2 days ago
RenewaBlox Research Paper
Why historical extrapolation is no longer sufficient and how RenewaBlox derives a defensible range
Jason Deane | Co-CEO RenewaBlox | July 2026
Executive summary
RenewaBlox models Bitcoin network hash rate from two primary drivers:
Net growth in electrical capacity actually allocated to mining and
Improvement in the effective efficiency of the global operating ASIC fleet.
Using a rounded mid July 2026 starting point of 900 EH/s, our constrained, base and accelerated cases produce July 2030 network hashrates of approximately 1,013 EH/s, 1,184 EH/s and 1,381 EH/s respectively.
These are scenarios, not predictions; their purpose is to make the assumptions behind our mining forecasts transparent and testable.
The old assumption
For much of Bitcoin’s history, assuming persistent double-digit hashrate growth was not especially controversial. New generations of ASIC hardware delivered large efficiency gains, capital flowed into industrial mining, and large amounts of underused or inexpensive power could be monetised by installing machines. An annual growth assumption of around 15–20% therefore looked conservative beside the historic chart. Indeed, RenewaBlox previously used much higher forecasts in the range of 17%-20% for its own original hashrate forecasts.
The problem is that a trend line describes what happened, but it does not explain why it happened. The market now contains structural forces that were either absent or far less important only a few years ago. Most notably, Bitcoin miners no longer have an uncontested claim on power-rich sites. AI and high-performance computing have created a competing source of demand for grid connections, substations, land and generation.
“Power is the scarce input in AI.” — Fred Thiel, Chairman and CEO of MARA, April 2026 [1]
That statement matters beyond AI. It captures the central change in mining economics: hardware can be ordered, but a suitable energised site may take years to develop and increasingly has several potential uses.
A model built from causes, not a trend line
Instead of selecting a network CAGR directly, our view is that it is derived from two variables.
The first is net mining-power growth. That is, the total of new megawatts energised for Bitcoin mining, less closures, conversions to other workloads and permanent retirements.
The second is effective fleet-efficiency improvement. This is the annual increase in hashes produced per megawatt by the average operating network fleet.
(1 + hashrate growth) = (1 + net mining-power growth) × (1 + fleet-efficiency improvement)
This structure avoids false precision. AI competition, grid queues, planning, capital availability, electricity prices, halvings and miner profitability are not separate deductions to be added together. They are influences on the amount of power that remains economically allocated to mining. Likewise, a new flagship ASIC does not instantly improve the whole network since machines are deployed gradually and older models continue running where power is cheap.
Scenario | Net mining-power growth | Effective fleet efficiency | Hashrate CAGR | July 2030 hashrate |
Constrained growth | -1.0% p.a. | +4.0% p.a. | 3.0% p.a. | ≈1,013 EH/s |
Base case | +2.0% p.a. | +5.0% p.a. | 7.1% p.a. | ≈1,184 EH/s |
Accelerated growth | +5.0% p.a. | +6.0% p.a. | 11.3% p.a. | ≈1,381 EH/s |
How the CAGR is calculated
The two percentages are multiplied, not simply added, because both effects occur at the same time: the network has more mining power and each megawatt produces more hashes.
(1 + hashrate growth) = (1 + net mining-power growth) × (1 + effective fleet-efficiency growth)
Worked calculations:
Scenario | Calculation | Resulting hashrate CAGR |
Constrained growth | 0.99 × 1.04 = 1.0296 | 2.96%, rounded to 3.0% |
Base case | 1.02 × 1.05 = 1.0710 | 7.10%, shown as 7.1% |
Accelerated growth | 1.05 × 1.06 = 1.1130 | 11.30%, shown as 11.3% |
For example, in the base case a 2% increase in mining power gives a multiplier of 1.02, while a 5% improvement in hashes per MW gives 1.05. Multiplying them gives 1.071, meaning the network hashrate is 7.1% higher after one year. The extra 0.1 percentage point is the compounding interaction between the two drivers.
These calculations produce annual hashrate growth rates of approximately 3.0%, 7.1% and 11.3%. We use a rounded 900 EH/s starting point because short-period estimates can move sharply with block luck and smoothing methodology.
For context, Hashrate Index showed a 30-day average around 970 EH/s in early June 2026, while its live pool page later displayed approximately 879 EH/s; this volatility is precisely why a rounded modelling anchor is preferable to treating one daily observation as an exact baseline. [2][3]
Why net mining-power growth may slow
AI has created a genuine alternative use for mining infrastructure
The strongest evidence is not a forecast model, it is the behaviour of mining companies themselves. Core Scientific, for example, agreed to modify existing sites to deliver approximately 200 MW of infrastructure for CoreWeave’s GPU workloads in 2024, later expanding contracted HPC infrastructure to approximately 590 MW across six sites. [4][5] MARA now describes a strategy covering Bitcoin mining, power generation, AI and HPC, while emphasising the value of sites with power, land, water and grid interconnection. [1][6]
“Our new contracts with CoreWeave position us to transform our hosting business and our earnings power by capturing exciting growth opportunities in AI compute.” — Adam Sullivan, CEO of Core Scientific, June 2024 [4]
This does not mean every Bitcoin mine can become an AI data centre. AI facilities require different connectivity, redundancy, cooling and construction standards. However, it does mean that suitable power infrastructure now carries an opportunity cost that did not exist at the same scale during the previous mining build-out.
The competing demand is large enough to matter
The International Energy Agency expects global data-centre electricity consumption to reach around 945 TWh by 2030, more than double its 2024 level. Its base case implies roughly 15% annual growth in data-centre electricity use from 2024 to 2030, with accelerated servers driven mainly by AI growing by about 30% a year. [7] In the United States, the IEA expects data centres to account for nearly half of electricity-demand growth to 2030. [8]
“Global electricity demand from data centres is set to more than double over the next five years.” — Fatih Birol, Executive Director of the IEA, April 2025 [8]
The implication for our model is not that AI directly subtracts a fixed number of percentage points from hashrate. Rather, it lowers the plausible net rate at which new, economically attractive megawatts remain dedicated to Bitcoin.
Why fleet efficiency still rises — but not at the headline rate
Frontier ASIC technology continues to improve. Bitmain announced the S19 Pro at 29.5 J/TH in 2020, the S21 Hydro at 16 J/TH in 2023, the S21 XP Hydro at 12 J/TH in 2024 and the S23 Hydro at as low as 9.5 J/TH in 2026. [9][10][11][12]
That chart could appear to support very high future hashrate growth but there is a distinction between new generation technology efficiency and effective fleet efficiency. While a newly announced machine may represent a substantial improvement over a four- or six-year-old unit, only a fraction of the global fleet is replaced each year. Deployment dates lag release dates, supply is finite, capital has a cost and older machines remain viable on unusually cheap or otherwise stranded energy.
Cambridge’s 2026 methodology explicitly distinguishes a machine’s release date from its deployment date and models the changing mix of profitable and unprofitable hardware. Its profitability framework also shows why the fleet becomes more efficient over time: as economics tighten, the minimum viable efficiency threshold falls and older equipment exits. [13][14]
Finally, the rate of improvement by generation of miner has slowed in relative terms as the physical limits of current chip development are reached.
Our 4–6% annual effective-fleet assumption therefore allows substantial technological progress without assuming that the entire network instantly adopts each flagship generation.
The halving makes a smooth CAGR unlikely
A four-year CAGR is useful in a financial model, but the real path is unlikely to be smooth. The 2028 halving will cut the block subsidy again and, unless offset by Bitcoin price, transaction fees or lower operating costs, this will compress miner revenue per unit of hashrate and accelerate the retirement of marginal machines. This is, in fact, the same behaviour as we have observed in previous halving events.
Since the mining industry has reached, in our view, somewhat of a maturation stage, it is possible that the effect of the halving in 2028 may have more impact this time round. It may be, for example, that marginal miners simply choose to exit the industry completely at that point rather than renew or upgrade their equipment. Instead, they may elect to repurpose their power contracts to AI or other HPC and, since the date of the halving can be predicted relatively accurately, this can be effectively planned for now.
The changing dynamics of the industry
It has been RenewaBlox’s firm belief for some time that the era of large industrial and centralised miners is coming to an end. With AI offering an alternative – and currently superior – revenue stream for large infrastructure operators with advantageous PPAs, the case for Bitcoin-only compute is no longer as clear cut. AI has a very high predicted growth curve and requires very significant amounts of capital as well as the full intent of a commercial organisation to focus, commit and deliver on it. In our view, we consider it unlikely that miner deployment on this scale will occur again.
However, we also believe that the industry is entering a new and more exciting phase; that of far more decentralised or utility-based mining operations. These are operations that simultaneously serve multiple purposes, such as grid balancing, heat provision, monetisation of wasted, stranded or curtailed energy (especially where flexibility is needed), improvement of the economics of existing or proposed renewable projects, mitigating methane on landfill or oil and gas sites, or simply adapting mining to solve a specific power problem in a specific way.
We consider this to be the main growth area for the industry over the next five years.
What the scenarios mean for a miner
Unlike a conventional sales forecast, higher network growth is adverse to an individual miner. For a fixed installed hashrate, Bitcoin production is broadly proportional to the miner’s share of the network. By July 2030, a static fleet would retain approximately 88.8% of its current network share in the constrained case, 76.0% in the base case and 65.2% in the accelerated case, before allowing for halvings, fees, uptime and pool effects.
Scenario | 2030 network | Share of current network position retained | Reduction in relative BTC output* |
Constrained | 1,013 EH/s | 88.8% | 11.2% |
Base | 1,184 EH/s | 76.0% | 24.0% |
Accelerated | 1,381 EH/s | 65.2% | 34.8% |
*Hashrate effect only; excludes halvings, transaction fees, uptime, curtailment and changes in the miner’s own fleet.
Conclusion
No one can know the network hashrate in July 2030 and the current transition in the mining industry reduces clarity. The purpose of the model is to replace an opaque assumption with a transparent chain of reasoning.
Our base case assumes that Bitcoin-dedicated power continues to expand, but more slowly than in the last cycle because grid-connected infrastructure has become scarcer and AI/HPC offers an alternative use. It also assumes that the average ASIC fleet continues to become materially more efficient, but at a lower rate than the improvement advertised for the newest machine. Those assumptions combine to produce 7.1% annual network growth and approximately 1.18 ZH/s by July 2030.
The constrained and accelerated cases then allow investors to test what happens if those structural changes prove more or less powerful than expected. That is the standard we believe a long-term mining forecast should meet: not certainty, but an explainable methodology, visible inputs and a range wide enough to expose the economics to genuine stress.
References
1. MARA, “MARA Advances Its Optimized Digital Infrastructure Strategy with Agreement to Acquire Long Ridge Energy Power”, 30 April 2026. https://ir.mara.com/news-events/press-releases/detail/1419/mara-advances-its-optimized-digital-infrastructure-strategy-with-agreement-to-acquire-long-ridge-energy-power
2. Hashrate Index, “Hashrate Index Roundup (June 8, 2026)”, 8 June 2026. https://hashrateindex.com/blog/hashrate-index-roundup-june-8-2026/
3. Hashrate Index, “Bitcoin Mining Pool Data”, accessed July 2026. https://hashrateindex.com/hashrate/pools
4. Core Scientific, “Core Scientific to Provide Approximately 200 MW of Infrastructure to Host CoreWeave’s HPC Services”, 3 June 2024. https://investors.corescientific.com/news-events/press-releases/detail/74/
5. Core Scientific, “Core Scientific and CoreWeave Announce $1.2 Billion Expansion at Denton, TX Site”, 26 February 2025. https://investors.corescientific.com/news-events/press-releases/detail/110/
6. MARA, “MARA 2025 Annual Report / Strategic Focus”, 2026 filing. https://ir.mara.com/sec-filings/all-sec-filings/content/0001507605-26-000007/0001507605-26-000007.pdf
7. International Energy Agency, “Energy demand from AI”, 2025. https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai
8. International Energy Agency, “AI is set to drive surging electricity demand from data centres”, 10 April 2025. https://www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-potential-to-transform-how-the-energy-sector-works
9. Bitmain, “ANTMINER S19 Pro launch information”, 2020. https://www.bitmain.com/news-detail/199
10. Bitmain, “ANTMINER S21 Hydro launch”, 9 October 2023. https://www.bitmain.com/news-detail/bitmain-releases-antminer-s21-hyd--s21-with-an-outstanding-energy-efficiency-of-16-jt-300
11. Bitmain Support, “S21 XP Hydro specification”, 2 July 2024. https://support.bitmain.com/hc/en-us/articles/34523540504857-S21-XP-Hyd-Specification
12. Bitmain, “BITMAIN showcases ANTMINER S23 Hydro at Bitcoin 2026”, 8 May 2026. https://www.bitmain.com/news-detail/447
13. Cambridge Centre for Alternative Finance, “CBECI Change Log v1.7.0”, 20 January 2026. https://ccaf.io/cbeci/change_log
14. Cambridge Centre for Alternative Finance, “CBECI Methodology”, accessed July 2026. https://ccaf.io/cbnsi/cbeci/methodology
15. Hashrate Index, “Global Hashrate Heatmap Update: Q2 2026”, 6 April 2026. https://hashrateindex.com/blog/global-hashrate-heatmap-update-q2-2026/