# Artikel Terkait Data Center

Pusat Berita HTX menyediakan artikel terbaru dan analisis mendalam mengenai "Data Center", mencakup tren pasar, pembaruan proyek, perkembangan teknologi, dan kebijakan regulasi di industri kripto.

Bitcoin Mining Farms Are Becoming AI Factories

Bitcoin mines are transforming into AI factories. This shift is driven by the convergence of three key assets from the previous crypto cycle: infrastructure, talent, and capital. Crypto mining companies like Crusoe, CoreWeave, and Bitdeer are repurposing their core competency—securing power, land, and grid connections in remote locations—to build data centers for AI clients. These firms are signing multi-billion dollar, long-term contracts with companies like Anthropic, AWS, and Microsoft, as AI's demand for reliable, high-capacity compute surpasses the profitability of Bitcoin mining. Simultaneously, crypto entrepreneurs and engineers are applying their skills to new AI ventures. Examples include OpenSea's co-founder launching OpenRouter (an AI model aggregator), and former Coinbase engineers building Fal.ai (a generative media infrastructure platform). Their experience in building scalable, global software networks translates effectively to the AI space. Furthermore, capital accumulated during the crypto boom is now fueling AI. Figures like Jed McCaleb (co-founder of Ripple) funded Voltage Park, a large-scale GPU cloud provider. Notably, some crypto investments, like FTX's early bets on Anthropic and Cursor, have generated astronomical paper returns, demonstrating how high-risk crypto capital flowed into AI before it became mainstream. The transition is not just about repurposing hardware, but about redirecting critical resources—power infrastructure, distributed systems expertise, and venture funding—to the next technological frontier: artificial intelligence.

链捕手9j yang lalu

Bitcoin Mining Farms Are Becoming AI Factories

链捕手9j yang lalu

Mining Stocks Are Moving Further Away from Crypto

Title: Mining Stocks Are Drifting Away from Crypto Summary: Despite Bitcoin (BTC) falling approximately 46% over the past year, leading Bitcoin mining stocks (e.g., HUT, WULF, IREN) have surged significantly. This divergence stems from a fundamental shift in how the market values these companies. Their stock prices are no longer tied primarily to crypto prices, mining output, or hash rates. Instead, investors are now pricing them as AI infrastructure plays. Mining companies possess critical assets for AI data centers: pre-permitted land, grid-connected power capacity, and operational expertise for high-load facilities—resources facing severe shortages and long lead times for new entrants. For example, CleanSpark signed a 20-year, ~$6.6 billion infrastructure lease for an AI data center, while Marathon Digital acquired a project with up to 2 GW of planned power capacity. Analysts note a strong correlation between a mining company's market valuation and its contracted or potential AI power capacity in North America. CoinShares predicts that by year-end, up to 70% of revenue for listed miners could come from AI/HPC, compared to about 30% at the start of 2026. However, this re-rating introduces new risks: 1) Valuation volatility linked to the broader AI/semiconductor sector, 2) Potentially low baseline return rates (estimated at 4-5% for some firms), and 3) Execution risks including massive financing needs, regulatory permits, and tenant quality. This strategic pivot is also changing miner behavior. They are selling BTC holdings more aggressively to fund AI capex, meaning selling pressure may persist regardless of Bitcoin's price. Furthermore, once power and sites are locked into long-term AI contracts, they are unlikely to return to Bitcoin mining, potentially altering the network's hash rate dynamics long-term. In essence, mining firms are being valued for what they are becoming—AI infrastructure providers—rather than pure-play crypto miners.

链捕手07/17 10:20

Mining Stocks Are Moving Further Away from Crypto

链捕手07/17 10:20

How Long Will the Storage Boom Last?

Title: "How Long Can the Memory Boom Last?" (Author: Takashi Yunoue) Summary: The semiconductor industry, especially the memory market, is experiencing an unprecedented, explosive boom. Data (WSTS, 1991-2026) shows that while other categories like micros, logic, and analog grew steadily, memory shipments, particularly DRAM and NAND flash, have seen a near-vertical spike since 2024. Monthly memory shipments surged from ~$5.6B in 2016 to ~$63.3B in 2026, an 11x increase, with year-on-year growth reaching a staggering 285%. This dwarfs the previous peak of ~60% during the 2017-2018 memory bubble. The primary driver is not volume but a ~10x price surge for both DRAM and NAND, fueled by insatiable demand from AI data centers. Hyperscalers like Amazon, Google, Microsoft, and Meta are making massive capital investments (projected at $755B in 2026, ~36x growth since 2015), creating a "black hole" that absorbs GPUs, High Bandwidth Memory (HBM), and high-performance storage. This has diverted production capacity, causing severe shortages and price hikes for memory in consumer electronics (PCs, smartphones). While forecasts now predict the global semiconductor market will hit $1.5 trillion in 2026 and memory alone may surpass $1 trillion by 2027, the author warns this boom is unsustainable. Historical analysis of memory market growth rates over 35 years shows that periods of sustained annual growth have never exceeded five consecutive years, inevitably followed by a downturn due to the "silicon cycle" (demand surge → price rise → overinvestment → oversupply → price crash). Given the current boom started from a low in 2023/2024, a peak is expected by 2027-2028 at the latest. Furthermore, a fundamental rule applies: "The higher the peak, the deeper the valley." The unprecedented 285% growth peak suggests the subsequent recession could be the most severe in industry history. The author cautions that the current soaring stock prices and wealth creation in the memory sector are based on inflated expectations and urges companies to use this prosperous period to prepare practically for the inevitable downturn.

marsbit07/14 21:20

marsbit07/14 21:20

37.7°C Heatwave Paralyzes Cambridge University AI Supercomputer, 350 Research Projects Halted

Cambridge University's national AI supercomputer, "Dawn," was forced offline for over a week due to a record-breaking June heatwave in the UK, where temperatures reached 37.7°C. This failure of the cooling system halted more than 350 critical research projects reliant on the £300 million machine, which features 1024 Intel GPUs and liquid cooling. Impacted work included climate change modeling, Parkinson's disease drug screening, and AI-powered cancer vaccine research, highlighting the vulnerability of essential infrastructure to extreme weather. The incident mirrors similar heat-related outages at Google and Oracle data centers in the UK in 2022. Experts point to a growing challenge: as AI chips become more powerful and generate significantly more heat, data center cooling systems—often designed for historical climate norms—are increasingly struggling against the rising temperatures driven by climate change itself.

marsbit07/09 03:55

37.7°C Heatwave Paralyzes Cambridge University AI Supercomputer, 350 Research Projects Halted

marsbit07/09 03:55

US Power Grid Sounds Red Alert Amid Scorching Heat, Bitcoin Mining Takes the Blame?

In late June 2026, a "heat dome" pushed the PJM Interconnection, the largest U.S. power grid, to its limits, triggering emergency orders from the Energy Secretary. One order allowed temporary suspension of environmental limits for power plants, while another mandated large users like data centers and bitcoin mines to cut grid power within 15 minutes, relying on backup generation instead. Contrary to appearing as passive "power hogs," major bitcoin miners are active participants in grid "demand response" programs. By voluntarily shutting down operations during peak stress, they receive significant financial credits. For example, Riot Platforms earned $21.0 million in Q1 2026 from such programs, drastically reducing its net mining costs. The article argues that the primary driver of the grid strain and soaring electricity prices is not bitcoin mining, which accounts for only 0.6%–2.3% of U.S. consumption, but the explosive growth of AI data centers. PJM forecasts a 32 GW demand surge by 2030, with 30 GW coming from data centers. Recognizing this shift, leading mining companies are strategically pivoting. They are leveraging their existing power infrastructure and contracts to transition into high-performance computing (HPC) and AI data center hosting, positioning this as a more lucrative future business model than traditional mining alone.

marsbit07/08 13:47

US Power Grid Sounds Red Alert Amid Scorching Heat, Bitcoin Mining Takes the Blame?

marsbit07/08 13:47

Anthropic Drops $19 Billion to 'Sponsor' a Bitcoin Miner

On July 6, 2026, Bitcoin mining company TeraWulf (NASDAQ: WULF) saw its stock surge 15% pre-market following the announcement of a landmark 20-year agreement with AI giant Anthropic. The deal grants Anthropic 401 MW of IT load capacity at TeraWulf's "Justified Data" campus in Kentucky and is projected to generate approximately $19 billion in contracted revenue for the miner over its duration. On the same day, TeraWulf also sold its 50.1% stake in a Texas joint venture for about $530 million. This dual move signals a strategic pivot: divesting non-core assets to fund its fully-owned Kentucky project and shifting focus from cryptocurrency mining to becoming a specialized infrastructure provider for AI. The agreement highlights a key advantage for Bitcoin miners transitioning to AI: their pre-existing access to land and critical power grid capacity, which is becoming a major bottleneck for data center expansion. Unlike peers who operate AI clouds, TeraWulf is adopting a "landlord" model, leasing only the physical space and power for clients' own servers. However, a significant gap exists between the deal's announcement and revenue generation, with the first phase of the Kentucky site not operational until late 2027. The $19 billion figure represents a long-term bet on both TeraWulf's execution and Anthropic's financial durability in the capital-intensive AI race. The market's positive reaction reflects the growing value of fundamental infrastructure—secured land and reliable electricity—in the era of AI compute scarcity.

marsbit07/08 03:56

Anthropic Drops $19 Billion to 'Sponsor' a Bitcoin Miner

marsbit07/08 03:56

DeepSeek Secretly Builds AI Chip, Specializing in Inference, Project Started a Year Ago with No Public Recruitments

DeepSeek, the Chinese AI company known for its algorithmic models, is secretly developing its own AI chip to reduce dependence on Nvidia, according to a Reuters report. The chip is designed specifically for AI inference, not training, and the project began approximately a year ago. Currently in early stages, DeepSeek is reportedly in discussions with chip design firms, foundries, and memory suppliers. The company, historically focused on algorithmic efficiency, has been discreetly hiring chip design engineers without public job postings. This move aligns with a global trend where major AI model companies like OpenAI and Anthropic are also pursuing custom chip development. DeepSeek founder Liang Wenfeng previously highlighted chip shortages as a challenge. While the company initially trained models on Nvidia H800s and later adapted to Huawei's Ascend chips, it now seeks greater control over its hardware foundation. Designing a competitive AI chip is a significant challenge, requiring years and substantial investment with no guarantee of success. However, DeepSeek's efforts are backed by a recent major funding round of approximately 51 billion RMB (about $7.4 billion) raised in June 2026. The funds are designated for expanding data centers based on domestic chips, developing proprietary AI chips, and recruiting top global talent. Infrastructure plans are also advancing, with job postings for data center design engineers, including projects in locations like Ulanqab, Inner Mongolia. The company remains characteristically low-key, with sources speaking anonymously and no official comment from DeepSeek itself. Nevertheless, this initiative marks a strategic expansion from software algorithms into the hardware layer that powers its AI systems.

marsbit07/08 02:04

DeepSeek Secretly Builds AI Chip, Specializing in Inference, Project Started a Year Ago with No Public Recruitments

marsbit07/08 02:04

Meta Joins the 'Selling Shovels' Game, Zuckerberg: 'Models Can Be Slow, but GPUs Must Earn'

Meta pivots to selling AI infrastructure as internal model development lags. Facing setbacks like Gemini usage restrictions and slower-than-expected progress on its own AI agents, the company is reportedly launching "Meta Compute." This service would rent out its massive GPU capacity and data center resources—projected to reach over 10GW—to external clients. The move mirrors SpaceX's "neocloud" model, offering high-margin, flexible compute contracts. Meta also plans to host third-party models like Anthropic's Claude, creating a platform similar to Amazon Bedrock. This strategy provides immediate revenue while its in-house models (like the upcoming "Watermelon") continue development. Wall Street reacted positively, seeing it as a way to monetize heavy AI investments even if Meta's models aren't yet industry-leading.

marsbit07/06 01:19

Meta Joins the 'Selling Shovels' Game, Zuckerberg: 'Models Can Be Slow, but GPUs Must Earn'

marsbit07/06 01:19

Tidal Investment: We Remain Bullish on the AI Industry Chain, But the Reasons Have Changed

Tidal Investment remains optimistic about the AI industry chain, but the rationale has shifted. The market narrative has changed. While recent large-scale IPOs (e.g., SpaceX) and major fundraising plans by tech giants like Alphabet and Meta have caused some nervousness, this isn't a sign of an AI peak. The focus has moved from the initial question of AI's viability to the sustainability of massive investment cycles. The key players—primarily the major cloud providers—are not slowing down; their capital expenditure (Capex) guidance for 2026 has been increased across the board (e.g., Alphabet to $180B, Amazon to $200B). This investment cycle is proving resilient and difficult to stop. Unlike traditional hardware cycles, current AI Capex is distributed across multiple physical layers—computing, memory, networking, and critically, power infrastructure. Bottlenecks are shifting from chips to elements like electricity, transformers, and cooling systems, which have much longer lead times and cannot be easily pre-built like fiber optics during the dot-com bubble. Supply chain data (e.g., Eaton's 240% YoY data center orders) confirms this broad-based, project-driven expansion. Market concerns are acknowledged but viewed differently. First, while Capex growth currently outpaces revenue growth, raising ROI questions, this mirrors the early scaling phase of cloud computing itself. A change in view would require concrete signals like downward Capex revisions or missed AI product targets, which haven't materialized by mid-2026. Second, comparisons to the 2000 dot-com bust are flawed. That crash was driven by a massive, parallel oversupply of cheap capacity (fiber). The current cycle faces *supply constraints* in critical, capital-intensive physical infrastructure that cannot be overbuilt as easily. In conclusion, the wave of fundraising reflects the next, more complex act of the AI story. Physical bottlenecks and sustained high Capex plans suggest this is not the finale but an ongoing, capital-intensive build-out phase. The script has changed, but the play is far from over.

marsbit06/25 10:36

Tidal Investment: We Remain Bullish on the AI Industry Chain, But the Reasons Have Changed

marsbit06/25 10:36

Stock Price Hits New High! Corning Targets CPO Optical Interconnect Market

TL;DR Corning's stock recently hit a new high, driven partly by investor interest in its role within AI data center optical interconnects. At a recent conference, Corning showcased its GlassBridge technology, a glass-based optical bridge designed to connect fibers to photonic integrated chips (PICs). This component targets challenges in Co-Packaged Optics (CPO) architectures by aiming to reduce coupling loss and simplify the precise alignment needed between optical fibers and nanoscale chip waveguides. The core issue GlassBridge addresses is the significant dimensional mismatch between fibers and PICs, which causes signal loss and assembly complexity. Corning leverages its glass and fiber expertise to create a platform it claims offers low coupling loss (demonstrating 1.5dB) and supports passive alignment. The broader vision extends to glass-based substrates with through-glass vias for advanced CPO packaging. While positioning itself as a solutions provider from cables to chip-level interconnects, Corning's GlassBridge is currently a technological showcase. Its path to widespread adoption in AI servers depends on overcoming hurdles in mass production yield, cost, thermal management, and crucially, validation and integration into future customer platforms from major cloud and chip companies competing with alternative solutions like silicon photonics.

marsbit06/25 03:20

Stock Price Hits New High! Corning Targets CPO Optical Interconnect Market