股票频道更多独家策划、专家专栏，免费查阅>> 责任编辑：栎树 中信证券认为太空算力的能源供给将高度依赖光伏技术。钙钛矿作为兼具高效率和低成本的下一代光伏 技术，有望成为重要的太空能源解决方案。2025年以来，钙钛矿产业化进展加速，效率、稳定性纷纷取 得突破，行业内多条GW级产线相继投产，建议关注钙钛矿产业投资机会。 据报道，太空AI中心是以AI算力太空部署为特征的新一代AI基础设施，旨在为AI发展突破能源、环 境、空间瓶颈。目前，全球科技力量争相布局太空AI中心建设，纷纷出台技术发展和建设规划，带动 相关技术和应用进入高速发展期。 太空算力正从概念验证走向可运行系统。海外方面，英伟达已完成H100GPU在轨测试，Space X计划4 年内每年完成部署100GW太空数据中心，谷歌拟2027年发射TPU原型卫星，Starcloud推进40MW数据中 心落地。中国数据中心"上天"进展顺利，已成功建起全球首个商业化AI卫星网络。2025年5月，中国国 星宇航与之江实验室联合发射了首批12颗、搭载AI芯片和模型的智能计算卫星，总算力达5 POPS，标 志着太空算力首次进入实际应用阶段。 ...

The chipmaker rallied during the second half of 2025, but can it sustain its momentum in 2026?When it comes to the artificial intelligence (AI) industry, no company is watched as closely as Nvidia (NVDA 0.44%). Over the last three years, the company's graphics processing units (GPUs) and its CUDA computing platform have become central pillars supporting generative AI development.Nvidia has become so influential in the AI realm that its quarterly earnings reports have essentially turned into a yardstick for ...

VANCOUVER, BC, Jan. 19, 2026 /PRNewswire/ -- Bitzero Holdings Inc., (CSE: BITZ) (OTC PINK: BTZRF) (FSE: 000) ("Bitzero" or the "Company"), a provider of sustainable blockchain and high-performance compute (HPC) data center infrastructure, today announced the acquisition of eight NVIDIA Blackwell B300 servers, totaling 64 GPUs, to be deployed at its Namsskogan, Norway facility in partnership with Hydra Host. The deployment, expected to be completed in Q1 2026, represents Bitzero's first direct investment in ...

Artificial intelligence is set to have another huge year.2026 is shaping up to be another fantastic year for the stock market, especially with the amount of artificial intelligence (AI) spending going on. I think there are several stocks bound to rocket higher over the year, and if you don't buy them now, you're going to regret it down the road.The five stocks on my shopping list for 2026 are Nvidia (NVDA 0.29%), Broadcom (AVGO +2.63%), Advanced Micro Devices (AMD +1.79%), Amazon (AMZN +0.49%), and Alphabet ...

Core Insights - The article discusses the evolution and significance of "pick-and-shovel" markets, highlighting how companies like Intel, Qualcomm, NVIDIA, and Amazon have adapted to changing technological landscapes and consumer demands [1][4][25]. Group 1: Historical Context - The concept of pick-and-shovel markets has existed for centuries, with Samuel Brennan being an early example of exploiting such a market during the gold rush [2][3]. - Brennan's strategy involved purchasing mining equipment and reselling it at high markups, leading to his success as the first West Coast millionaire [3]. Group 2: Modern Examples - Intel became a dominant player in the CPU market after being selected as the microprocessor supplier for IBM PCs, demonstrating the effectiveness of scale in pick-and-shovel businesses [4][5]. - Qualcomm and NVIDIA also exemplify this trend, with Qualcomm leading in digital communications chipsets and NVIDIA dominating the GPU market in data centers [5][6]. Group 3: Innovation and Scale - The success of these companies is attributed to a combination of foresight, skill, and strategic partnerships, which allowed them to leverage their scale for competitive advantage [6][7]. - All three companies utilize backward-compatible designs to guide buyers' upgrade paths, preventing rivals from gaining similar advantages [8]. Group 4: Amazon's Mechanical Turk - Amazon's Mechanical Turk (MTurk) emerged as a solution to product labeling challenges, allowing users to perform microtasks that machines could not handle effectively [12][15]. - Despite initial challenges, MTurk eventually demonstrated its scalability, particularly when used by researchers like Fei-Fei Li for large-scale image labeling [19][20]. Group 5: Economic Implications - The article emphasizes that while investments in innovation are beneficial for society, they can also lead to monopolistic behaviors, as seen in the historical context of pick-and-shovel markets [28][29]. - Companies in these markets share similar incentives to innovate, but the outcomes can vary, as illustrated by Amazon's experience with MTurk, which did not yield significant financial returns despite its impact [26][27].

Submitted by Tight SpreadsThe DOE has been flooding their sites with white-papers and latest Fusion Science and Technology Roadmap (FS&T Roadmap) are not just a scientific plan, but an industrial policy designed to transition fusion from the laboratory to the commercial market by the mid-2030s. The DOE has prioritized its actions to align with the aggressive “fast-track” development cycles of nuclear fission and fusion companies:Near-Term (Next 2-3 Years): Digitalization & Infrastructure PrepAI-Fusion Conve ...

Advanced Micro Devices (NASDAQ: AMD stock) projects that if its ambitious data-center expansion targets materialize, shareholders could see massive upside through 2030. The claim has Wall Street divided between believers who see generational growth and skeptics who question whether the valuation already prices in much of that upside. Management's growth case: Can AMD sustain 60% data center CAGR? Copy link to section At its Financial Analyst Day in November 2025, AMD laid out a roadmap. The company expects ...

Quantum computing is the next big thing in tech, but you don't need to invest in any risky start-ups to profit from it.Quantum computing used to be one of those things you only read about in a science fiction novel, along with faster-than-light travel or teleportation. It might be part of everyday talk on the Starship Enterprise, but it's not real, right?Turns out it's very real, and it's one of the fastest-growing industries out there. McKinsey estimates the quantum computing market could grow from about $ ...

科技巨头英伟达去年5月发的一份技术报告，本来想推广他们的新高压直流系统，结果里面藏了个大 bug。报告说，一个1吉瓦的数据中心机架母线得用掉50万吨铜。这数字听起来吓人，相当于全球铜供应 量的1.7%。 大家一看，人工智能数据中心这么耗资源，赶紧传开了。分析师们没多想，就拿这个当证据，写报告说 人工智能会把铜吃光。华尔街的投行们在6月的商品展望里直接引用，预测如果建多个这样的中心，铜 需求得占全球产量大头。国内券商也跟风，出一堆专题，说人工智能扩张会引发铜短缺。媒体一报道， 投资者就激动了。 这份报告本来是技术性的，焦点在800V HVDC架构上，能降低能耗。但那个铜用量计算明显不对劲。 报告里说每兆瓦机架用200公斤铜，1吉瓦是1000兆瓦，算下来该是200吨。可他们写成50万吨，差了 2500倍。起初没人查，报告被当成金科玉律。技术博客先转载，强调人工智能对资源的压力。 专业会议上，能源专家们讨论时，总拿这个举例，说技术进步得付出代价。几个月下来，研究机构像 ThunderSaid Energy虽觉得怪，但没马上说破。相反，其他公司出报告，基于这个推演到2030年，人工 智能集群得要几百万吨铜。基金经理一 ...

1月19日，"铜博士"行情明显降温。沪铜期货主力合约跌0.68%，江西铜业、云南铜业、铜陵有色等龙头 股纷纷回调。 市场调整与英伟达的数据乌龙不无关系。英伟达此前发布的技术论文称，1吉瓦（GW）规模的传统数 据中心机架需要"50万吨"铜用于母线，近日修正为约200吨，修正后数值仅为原数据的0.04%。 这一数据修正对铜价究竟影响几何？ 英伟达数据乌龙持续发酵 1月19日，A股铜概念股在经历前期强势上涨后集体回调。截至收盘，Wind铜产业指数跌0.88%，江西铜 业跌5.63%，云南铜业跌3.74%，铜陵有色跌3.27%。 期货市场同步承压。截至收盘，沪铜期货主力合约收于101180元/吨，跌0.68%，延续回调态势。现货铜 价虽维持高位，但伦敦金属交易所（LME）与沪铜期货价格短期承压，叠加库存回升与下游观望情 绪，市场进入阶段性平衡。 铜价此番调整与英伟达一份技术论文的数据乌龙不无关系。英伟达在2025年5月发布的论文称，1吉瓦 （GW）规模的传统数据中心机架需要"50万吨"铜用于母线。能源咨询机构Thunder Said Energy表示， 英伟达很可能是将"50万磅"（约227吨） 误写成"50万吨"， ...