A History of Technological Evolution Powered by Electricity: Aluminum, Bitcoin, and AI

marsbitPublished on 2026-05-26Last updated on 2026-05-26

Abstract

The journey from the Rockdale aluminum smelter in Texas to space-based data centers illustrates a core economic principle: whoever controls the cheapest electricity dictates the use of computing power. The evolution is clear. Old industrial sites with pre-existing, high-capacity power grids are being repurposed. In Rockdale, a former Alcoa plant now houses vast Bitcoin mining rigs, which are increasingly being replaced by AMD chips for AI training. The logic is purely financial: while smelting aluminum yields $0.17–0.27 per kWh and Bitcoin mining $0.05–0.11, AI inference on H100 GPUs generates $1.27–3.67 per kWh. Recent deals confirm the rush for power infrastructure. Riot Platforms leases space to AMD; TeraWulf bought an old Kentucky aluminum plant for its grid; NYDIG secured a New York site for its cheap hydropower to mine Bitcoin. As AI giants like Anthropic, Microsoft, Google, and Amazon aggressively expand, they now directly compete with crypto miners for the same industrial power resources, often outbidding them. This has led to a decline in Bitcoin's global hash rate and a wave of miner conversions to AI data centers. This "digital resource curse" extends globally. Gulf nations, long offering subsidized power to attract heavy industry like aluminum, are now pivoting to become AI and cloud computing hubs—exporting computational power instead of physical commodities. Similarly, Bhutan halted its sovereign Bitcoin mining to sell hydropower directly to India for a stead...

Written by: Thejaswini M A

Compiled by: Luffy, Foresight News

Drive about an hour northeast from Austin, Texas, past barbecue joints and barren scrubland, and you'll arrive in Rockdale, Texas. If you roll down the window before the town's outline comes into view, you'll hear a steady roar, like the rumble of a jet engine idling on the tarmac.

Rockdale, built around an old aluminum smelter, now hosts North America's largest concentration of Bitcoin mining operations, with industry leaders like Riot Platforms and Bitdeer having set up shop there. Investigations by The New York Times and Al Jazeera have long documented this soundscape: tens of thousands of mining rigs paired with industrial fans running at full throttle to prevent equipment from overheating and shutting down in the Texas heat.

Follow the roar into the former plant of Alcoa, a symbol of 20th-century heavy industry. No trace of aluminum production remains in the vast industrial yard. Inside a massive metal warehouse, countless thick copper cables and industrial server racks are arranged in grids. Computer equipment is fully submerged in churning synthetic coolant.

Originally used for Bitcoin mining, the equipment here is now being gradually replaced with AMD chips, transitioning to handle artificial intelligence model training workloads.

Forget debating whether AI is a bubble or Bitcoin is in terminal decline. This industrial succession is just a surface manifestation. The companies securing leases for these sites see it clearly: the true core asset is the electrical infrastructure. This is now an industry-wide consensus.

If you're still wondering why, the underlying logic stems from the disparity in revenue per kilowatt-hour of electricity (calculated at real-time London Metal Exchange prices):

  • Aluminum Smelting: Each kWh can generate $0.17–$0.27 in gross revenue.
  • Bitcoin Mining: At current rates, each kWh yields only $0.05–$0.11.
  • Running AI Inference Tasks on H100 GPUs: Each kWh yields a whopping $1.27–$3.67.

When electricity was cheap, producing aluminum made sense. As margins in the aluminum industry were squeezed, Bitcoin mining became the new use case for low-cost power. By 2026, with Bitcoin prices depressed, AI workloads have become the unequivocally better choice.

Three recent transactions vividly illustrate the industry-wide scramble for power resources, whether for cryptocurrency mining or AI compute.

Riot, with its large site in Rockdale, hasn't limited itself to Bitcoin mining. It has leased part of its space and power to chip giant AMD for building an AI data center. Simply by subleasing electricity and space, the company stands to earn hundreds of millions of dollars.

TeraWulf has embarked on a major expansion, spending $200 million to acquire the century-old Century Aluminum plant in Hawesville, Kentucky. The core reason for choosing this location is the plant's existing, robust high-capacity electrical infrastructure. The company plans to dismantle the old production equipment and build a large-scale data center campus leveraging the ready-made power grid.

NYDIG set its sights on the former Massena East industrial site in New York State. The site had been idle for years but offers direct access to 435 megawatts of cheap hydroelectric power from the St. Lawrence River. While peers pivot to AI, NYDIG secured this location solely to lock in low-cost hydropower to continue its Bitcoin mining operations. The industry has long since stopped building sites from scratch; it now competes for existing power hubs.

Over the past two decades, Bitcoin miners have roamed the globe in search of cheap power: at remote hydroelectric dams in Washington state, at flare gas sites in North Dakota's oil fields, in clusters of old industrial grids in upstate New York. The industry has honed a mature set of supporting capabilities: 24/7 high-load power management, industrial-grade cooling solutions, long-term, low-cost power contracts.

The rising AI firms happen to need these ready-made resources and have deeper pockets.

Anthropic is aggressively locking down power resources. Microsoft, Google, and Amazon are also on a data center construction spree, with power infrastructure buildout struggling to keep pace with new facility deployment. The three tech giants are now directly competing with Bitcoin miners for the same pool of industrial power resources. Where miners once competed against each other for power, they now find themselves at a disadvantage against the tech titans.

Data from early 2026 confirms the industry's predicament. Bitcoin's global network hash rate declined for the first time in six years. While the current cost to mine a single Bitcoin is around $88,000, for most of May this year, the price has hovered around $77,000. Miners operating at standard electricity rates are losing money on every coin they produce.

A collective pivot has followed. Companies like Hive, Hut 8, TeraWulf, and Iren are dismantling mining rigs and converting facilities into AI server rooms. CoreWeave has completely exited crypto mining, fully transitioning to AI cloud services. MARA acquired a French tech company to shift its business focus. Those positioned as "power operators," holding valuable electricity resources, are surviving. Miners solely focused on the crypto space are in crisis.

Energy analysts have dubbed this phenomenon the "digital resource curse": nations and companies are increasingly realizing that simply controlling cheap electricity yields higher returns than developing new technologies themselves.

The Gulf States grasped this logic early. For the past six decades, they have maintained ultra-low electricity tariffs: Kuwait has kept residential rates stable at $0.007 per kWh since 1966; in Abu Dhabi, the combined cost of production and delivery is about $0.087 per kWh, but it's sold to residents for just $0.014. Low-cost power was originally a tool for industrial policy, attracting energy-intensive industries like aluminum smelting, petrochemicals, and steel to the desert.

Today, that same cheap power that once supported heavy industry has found a new user: the data center. Saudi Arabia has established a sovereign AI investment agency, HUMAIN, investing tens of billions in tech infrastructure. The UAE is building a 5-gigawatt AI park, attracting companies like OpenAI, Oracle, and Nvidia. The grid that once powered aluminum smelters now powers AI compute. The planned NEOM Oxagon, originally conceived as a floating industrial city, has been repurposed into a $5 billion AI data center cluster powered by wind and solar.

The Carnegie Endowment for International Peace noted: Cloud computing has become the Gulf's "new aluminum." Instead of exporting physical commodities, the region is now converting fossil fuels and solar energy into compute power and exporting it via the internet.

It's not just the Middle East. Bhutan's case is equally telling.

Bhutan once possessed some of the world's cheapest hydropower. Its state-backed Bitcoin mining project was once hailed as a model of sovereign mining, holding a peak of 13,000 BTC. That has now dwindled to 3,100 BTC, and mining operations ceased over a year ago. The country's hydropower is now sold directly to the Indian grid.

The calculus mirrors that of the old American aluminum plant: Is Bitcoin mining still the best use of our electricity? When the answer was yes, Bhutan mined. When selling power to India offered a more stable return without the volatility risk of cryptocurrency, the power flowed to its neighbor.

Similarly, Starcloud raised $200 million to plan an orbital solar-powered data center. They recently trained the first AI model in space using H100 GPUs and are seeking approval to launch 88,000 satellites. While the project retains a Bitcoin mining operation, it is only a secondary component: the in-orbit solar panels generate continuous power, and when the AI compute queue is idle, the surplus electricity is used for crypto mining.

Low Earth orbit offers unique advantages for power generation: constant sunlight, no land use, and the cold of space negates the need for extensive cooling equipment. Over the past two decades, launch costs have fallen by 95%.

SpaceX is also deeply involved in the power-and-compute calculus. According to its latest IPO filing, the company's Colossus 1 data center in Memphis, Tennessee, has its entire compute capacity exclusively leased by Anthropic under a contract running until May 2029, valued at over $40 billion. This single deal alone brings SpaceX $1.25 billion in monthly revenue. This data center was also converted from an old appliance factory, mirroring the transformation model seen in Rockdale's aluminum plant.

Amidst all this industrial churn, perhaps the most unexpected pivot is that of Allbirds. The sustainable shoe brand, once valued at $4 billion, saw its stock price plummet 98% as the consumer brand bubble burst. With its core business struggling, the company, armed with cash and a public listing, decisively pivoted entirely to operating AI compute infrastructure. Its stock price subsequently surged 350%. The market voted: operating servers and trading in power/compute is far more profitable today than the traditional consumer sector.

Meanwhile, crypto projects like Bittensor, Render, and Akash are charting a different path: not building large, centralized facilities, but aggregating the world's distributed, idle compute.

Bittensor has built a compute marketplace where, underpinned by a fixed-supply token system, various AI models compete to provide answers; the project also halved its daily token issuance in December 2025. Render incentivizes users to share idle GPU resources for AI tasks. Akash rents out cloud compute, claiming prices up to 85% lower than Amazon Web Services.

This distributed compute model is gaining attention. At Nvidia's GTC conference in 2026, CEO Jensen Huang compared Bittensor to the classic internet project Folding@home. The latter was born from the idea of leveraging idle home computers worldwide, making dormant devices productive; Bittensor uses crypto tokens as an incentive to aggregate the compute of idle gaming PCs and old mining rigs.

Looking at the big picture, from the roar of industrial fans in Rockdale to sun-chasing satellites in orbit, a massive reconfiguration around physical assets is underway. For the companies involved, the only rule is to chase profit margins. I predict that a decade from now, these compute facilities may once again be emptied and repurposed for the next generation of emerging industries. The underground electrical transmission lines, however, will remain unchanged from beginning to end.

Whoever controls the cheapest electricity decides what the compute is used for. This logic has already played out in Texas, Bhutan, and Abu Dhabi. And 250 miles above the Earth's surface, it will hold true as well.

Related Questions

QWhat is the core reason behind the industrial transformation from aluminum smelting to Bitcoin mining and then to AI data centers at sites like the one in Rockdale, Texas?

AThe core reason is the pursuit of maximizing revenue per kilowatt-hour of electricity consumed. The decision is driven by comparing the gross revenue each activity generates from the same unit of power: aluminum smelting ($0.17–0.27/kWh), Bitcoin mining ($0.05–0.11/kWh), and AI inference tasks using H100 GPUs ($1.27–3.67/kWh). As profitability shifts, businesses repurpose infrastructure to host the most lucrative power-consuming operation.

QHow are Bitcoin miners being outcompeted in the race for cheap electricity resources according to the article?

ABitcoin miners are being outcompeted by well-funded AI companies and tech giants like Microsoft, Google, and Amazon. These AI-focused firms have stronger capital and are aggressively securing the same industrial-scale, low-cost power resources and data center sites that miners previously relied on. This competition has contributed to a decline in Bitcoin's global hash rate as miners struggle with profitability against higher electricity bidders.

QWhat is the 'digital resource curse' mentioned in the article, and how is it exemplified by Gulf states?

AThe 'digital resource curse' refers to the phenomenon where countries or companies find that simply controlling cheap electricity resources yields higher returns than developing new technologies. Gulf states like Kuwait and Abu Dhabi exemplify this by using long-standing, heavily subsidized low electricity prices—initially to attract energy-intensive industries like aluminum and chemicals—to now pivot and attract massive AI data center investments, effectively exporting compute power instead of physical commodities.

QWhat alternative, decentralized model for providing AI compute is highlighted in the article, and which projects represent it?

AThe article highlights a decentralized model that aggregates globally distributed, idle computing power instead of building large centralized data centers. This model is represented by crypto projects like Bittensor (which creates a marketplace for AI models to compete), Render (which shares idle GPU resources for AI tasks), and Akash (which offers decentralized cloud computing at lower costs). Nvidia's CEO compared Bittensor to the classic distributed computing project Folding@home.

QWhat broader principle does the article conclude governs the evolution of power-intensive industries from aluminum to AI?

AThe article concludes that the fundamental principle is that whoever controls the cheapest electricity determines the use of compute power. Industries—whether aluminum smelting, Bitcoin mining, or AI training—will continuously migrate to and repurpose infrastructure based on where they can achieve the highest profit per unit of electricity. The physical power grid remains constant, while the operations it supports evolve to follow the highest monetary return.

Related Reads

Just Now, Chinese AI Enters Top 2 in Global Programming, Only Claude Remains Ahead

**China's AI Ranks Second Globally in Programming, Trailing Only Claude** Today, Alibaba's Qwen3.7-Max achieved a score of 1541 on the Code Arena benchmark, securing fourth place globally and surpassing top models like GPT-5.5 and Gemini 3.5 Flash. Among the top positions, it is now the only non-Claude model, placing second overall after Anthropic's Opus models. Before this official ranking, Qwen3.7-Max had already gained recognition overseas. In practical tests, it outperformed rivals on tasks like creating a self-training Tetris AI and generating complex 3D models, often at a significantly lower cost. Developers praised its ability, especially when integrated with tools like Hermes Agent and OpenCode, to effectively replace models such as GPT-5.5. In a hands-on challenge to create a 3D racing game from a detailed prompt, Qwen3.7-Max delivered a fully playable HTML file in the first attempt, requiring only minor bug fixes. It uniquely included a start menu and sound effects—details missed by other models. While competitors like Gemini 3.5 Flash and Claude Opus 4.6 produced less polished or functional versions, and GPT-5.5 had its own quirks, Qwen3.7-Max stood out for its initial completeness and playability. This performance stems from its design as an "Agent Base Model," built for long-duration, autonomous task execution. Internal tests show it can run continuously for 35 hours, making over 1158 tool calls without context degradation or instruction drift. Key technical advancements include "environment expansion" training, which improves adaptability across different frameworks, and "long-horizon autonomous execution" training, enabling sustained strategic decision-making. By entering the top tier of the programming arena, Qwen3.7-Max demonstrates that Chinese AI models are not just catching up but are becoming defining competitors, challenging the long-standing dominance of Silicon Valley in this field.

marsbit20m ago

Just Now, Chinese AI Enters Top 2 in Global Programming, Only Claude Remains Ahead

marsbit20m ago

From a Lunch Table to an Infinite Universe: Fei-Fei Li Bets on AI's Next Dimension

From a Lunch Table Conversation to an Infinite Universe: Fei-Fei Li Bets on AI's Next Frontier - Spatial Intelligence In an era dominated by large language models, AI pioneer Fei-Fei Li argues that true understanding requires spatial intelligence — the ability to perceive, reason, and interact within the physical 3D/4D world. She points to evolutionary history: spatial perception drove the Cambrian explosion 540 million years ago, while language is a far more recent, inherently "lossy" way to encode reality. Current models struggle with basic spatial tasks a child can do, like counting chairs in a video. Her company, World Labs, is pioneering this shift with "Marble," a model that generates navigable, consistent 3D worlds from text, images, or simple 3D inputs—distinct from video generators like Sora. Though smaller than models like GPT-5, due to scarce 3D data and early-stage scaling laws, Marble is already used in gaming, robot training (by NVIDIA), architectural design, and personalized therapy for conditions like OCD and acrophobia. Li envisions this technology enabling "infinite universes" for creativity, social interaction, and more. However, she cautions against utopian or dystopian extremes, advocating for a measured vision where AI enhances human dignity and prosperity, akin to how electricity transformed civilization. The journey is long — as evidenced by the 20-year path to viable autonomous vehicles — but the direction is clear: for AI to move from merely talking about the world to truly understanding and acting within it.

marsbit23m ago

From a Lunch Table to an Infinite Universe: Fei-Fei Li Bets on AI's Next Dimension

marsbit23m ago

X Stock Market Investment & Trading: A 'Noise-Free' List of 50 Key Accounts

Titled "A 'Noise-Reduction' List of 50 Top US Stock Market Influencers on X," this article curates a selection of accounts for investors seeking quality information beyond follower counts. The list prioritizes accounts that have consistently discussed US stocks, ETFs, earnings, macroeconomics, options, and tech/AI/semiconductor topics over the past 90 days. Selection criteria focused on genuine informational value, stable analytical frameworks over mere news aggregation, and a clear relevance to US equity markets, with tighter filtering for crypto-heavy accounts. The final 50 accounts are categorized into three groups: - **Core US Stocks/Trading (31 accounts)**: Covering market trends, individual stocks, earnings, valuation, options, macroeconomics, and trading strategies. - **Tech/AI/Semiconductors (18 accounts)**: Focused on tech stocks, AI supply chains, semiconductor cycles, data centers, and cloud capital expenditure. - **News Source (1 account)**: Useful as a news radar, not for standalone decision-making. Presented alphabetically by handle, the list includes analysts, traders, and researchers such as @amy6tina (options/CFA), @dylan522p (semiconductors/AI infrastructure), @gerberkawasaki (tech stocks), @jimcramer (market commentary), and @tengyanai (semiconductors/AI trends). The article suggests using the list to: 1) complete one's information sources on US markets and specific sectors, 2) observe narrative linkages between AI, semiconductors, earnings, macro liquidity, and stock prices, and 3) "de-noise" and enhance the quality of one's X feed. It clarifies this is not investment advice or an endorsement, but a snapshot of content relevance and informational value for US equity investors.

marsbit29m ago

X Stock Market Investment & Trading: A 'Noise-Free' List of 50 Key Accounts

marsbit29m ago

Trading

Spot
Futures

Hot Articles

What is $BITCOIN

DIGITAL GOLD ($BITCOIN): A Comprehensive Analysis Introduction to DIGITAL GOLD ($BITCOIN) DIGITAL GOLD ($BITCOIN) is a blockchain-based project operating on the Solana network, which aims to combine the characteristics of traditional precious metals with the innovation of decentralized technologies. While it shares a name with Bitcoin, often referred to as “digital gold” due to its perception as a store of value, DIGITAL GOLD is a separate token designed to create a unique ecosystem within the Web3 landscape. Its goal is to position itself as a viable alternative digital asset, although specifics regarding its applications and functionalities are still developing. What is DIGITAL GOLD ($BITCOIN)? DIGITAL GOLD ($BITCOIN) is a cryptocurrency token explicitly designed for use on the Solana blockchain. In contrast to Bitcoin, which provides a widely recognized value storage role, this token appears to focus on broader applications and characteristics. Notable aspects include: Blockchain Infrastructure: The token is built on the Solana blockchain, known for its capacity to handle high-speed and low-cost transactions. Supply Dynamics: DIGITAL GOLD has a maximum supply capped at 100 quadrillion tokens (100P $BITCOIN), although details regarding its circulating supply are currently undisclosed. Utility: While precise functionalities are not explicitly outlined, there are indications that the token could be utilized for various applications, potentially involving decentralized applications (dApps) or asset tokenization strategies. Who is the Creator of DIGITAL GOLD ($BITCOIN)? At present, the identity of the creators and development team behind DIGITAL GOLD ($BITCOIN) remains unknown. This situation is typical among many innovative projects within the blockchain space, particularly those aligning with decentralized finance and meme coin phenomena. While such anonymity may foster a community-driven culture, it intensifies concerns about governance and accountability. Who are the Investors of DIGITAL GOLD ($BITCOIN)? The available information indicates that DIGITAL GOLD ($BITCOIN) does not have any known institutional backers or prominent venture capital investments. The project seems to operate on a peer-to-peer model focused on community support and adoption rather than traditional funding routes. Its activity and liquidity are primarily situated on decentralized exchanges (DEXs), such as PumpSwap, rather than established centralized trading platforms, further highlighting its grassroots approach. How DIGITAL GOLD ($BITCOIN) Works The operational mechanics of DIGITAL GOLD ($BITCOIN) can be elaborated on based on its blockchain design and network attributes: Consensus Mechanism: By leveraging Solana’s unique proof-of-history (PoH) combined with a proof-of-stake (PoS) model, the project ensures efficient transaction validation contributing to the network's high performance. Tokenomics: While specific deflationary mechanisms have not been extensively detailed, the vast maximum token supply implies that it may cater to microtransactions or niche use cases that are still to be defined. Interoperability: There exists the potential for integration with Solana’s broader ecosystem, including various decentralized finance (DeFi) platforms. However, the details regarding specific integrations remain unspecified. Timeline of Key Events Here is a timeline that highlights significant milestones concerning DIGITAL GOLD ($BITCOIN): 2023: The initial deployment of the token occurs on the Solana blockchain, marked by its contract address. 2024: DIGITAL GOLD gains visibility as it becomes available for trading on decentralized exchanges like PumpSwap, allowing users to trade it against SOL. 2025: The project witnesses sporadic trading activity and potential interest in community-led engagements, although no noteworthy partnerships or technical advancements have been documented as of yet. Critical Analysis Strengths Scalability: The underlying Solana infrastructure supports high transaction volumes, which could enhance the utility of $BITCOIN in various transaction scenarios. Accessibility: The potential low trading price per token could attract retail investors, facilitating wider participation due to fractional ownership opportunities. Risks Lack of Transparency: The absence of publicly known backers, developers, or an audit process may yield skepticism regarding the project's sustainability and trustworthiness. Market Volatility: The trading activity is heavily reliant on speculative behavior, which can result in significant price volatility and uncertainty for investors. Conclusion DIGITAL GOLD ($BITCOIN) emerges as an intriguing yet ambiguous project within the rapidly evolving Solana ecosystem. While it attempts to leverage the “digital gold” narrative, its departure from Bitcoin's established role as a store of value underscores the need for a clearer differentiation of its intended utility and governance structure. Future acceptance and adoption will likely depend on addressing the current opacity and defining its operational and economic strategies more explicitly. Note: This report encompasses synthesised information available as of October 2023, and developments may have transpired beyond the research period.

363 Total ViewsPublished 2025.05.13Updated 2025.05.13

What is $BITCOIN

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of BTC (BTC) are presented below.

活动图片