来自 Maksym Misichenko · Nasdaq ·
AI智能体对这条新闻的看法
The panel is bearish on Oklo's co-located SMR data-center concept, citing regulatory hurdles, scale economics, operational risks, and potential architectural mismatches with hyperscalers' redundancy requirements.
风险: Securing reliable fuel supply chains and waste disposal for scaled SMR deployments, as well as potential rejection by hyperscalers due to redundancy requirements.
机会: Potential timeline advantage through federal preemption powers for nuclear projects.
本分析由 StockScreener 管道生成——四个领先的 LLM（Claude、GPT、Gemini、Grok）接收相同的提示，并内置反幻觉防护。 阅读方法论 →
盖洛普民意调查显示，71%的美国人反对人工智能数据中心，原因是它们对公用事业成本和环境造成影响。
核能初创公司Oklo正在与电力和冷却系统设计公司Vertiv合作，共同建设小型模块化核反应堆和人工智能数据中心。
这种组合可以通过产生清洁能源和避免分区限制来规避地方反对。
人工智能数据中心建设刚刚遭遇了挫折。
根据一项新的盖洛普民意调查，71%的美国人反对在其所在地区建设任何人工智能数据中心，其中48%的人强烈反对。而只有7%的人强烈支持此类建设。这是一个令人难以置信的不平衡结果。
人工智能会创造世界上第一个万亿富翁吗？ 我们的团队刚刚发布了一份报告，内容是关于一家鲜为人知但提供英伟达和英特尔都迫切需要的关键技术公司，被称为“不可或缺的垄断”。继续 »
由于原因各异，但最常见的担忧是数据中心过度使用水和电，对环境的影响，以及它们可能提高公用事业账单的潜力。但为了使公司的AI投资获得回报，基础设施必须在没有陷入多年的诉讼或官僚程序的情况下得到建设。有没有办法使人工智能数据中心更令人接受？
出人意料的是，核能公司Oklo (纽约证券交易所: OKLO) 也许对这个问题有一个意想不到的解决方案。以下是这家充满活力的初创公司的雄心勃勃的商业计划如何让每个人都受益（包括其股东）。
对人工智能数据中心强烈的两党反对正在扰乱全国各地的AI建设。
地方反对已经阻止或推迟了自2024年以来至少640亿美元的美国数据中心建设项目，预计随着更多项目提交给地方分区委员会和其他对选民负责的决策者，这一数字还会增加。例如，弗吉尼亚州沃伦顿镇的愤怒居民投票将所有支持建设当地亚马逊数据中心的城镇议员赶出。新的委员会一致投票批准了一项改变分区规则的措施，永久禁止整个城镇建设所有数据中心。
类似的故事最近也在俄勒冈州、犹他州、新泽西州和马萨诸塞州上演。本月，缅因州成为第一个通过州范围内禁止人工智能数据中心建设的州。那么，一家核能初创公司如何能够阻止这股潮流呢？
Oklo是一家尚未商业化的公司，目前正在与爱达荷国家实验室合作建设其第一个原型动力装置，配备小型模块化反应堆（SMR）。建成后，这个“奥罗拉动力站”将提供高达75兆瓦的核能——远低于大型核电站，但足以供大约50,000户家庭使用。
由于人工智能数据中心和核电站都需要巨大的、耗电量大的冷却系统，Oklo与制造商Vertiv (纽约证券交易所: VRT) 合作开发了一种用于安置Oklo SMR和人工智能数据中心的设施的冷却系统。相同的系统将冷却反应堆和数据中心，反应堆将提供所有所需电力。
这种组合的SMR/数据中心不仅能更节能，还可以远离现有的电网（以及市政当局令人讨厌的分区法规）。或者，它可以将其多余的电力输回当地电网，从而缓解潜在的能源短缺。Oklo计划在其第一个奥罗拉动力站展示这种组合的冷却系统技术，该动力站计划在明年晚些时候投入运营。
Oklo和Vertiv尚未公开披露共建的SMR数据中心将是什么样子，或者它有多大、有多强大。而且，除非Oklo开始将奥罗拉动力站投入运营，否则我们不会知道这是否是一个可行的计划，更不用说一个可以全国规模推广的计划了。
不过，盖洛普民意调查显示，核电站建设的民意支持率比人工智能数据中心建设高出18个百分点。它还发现，无碳核能是最不具有政治分歧性的能源形式。因此，即使需要数年时间才能将Oklo核能供电的人工智能数据中心推向市场，对于目前看到其数据中心陷入昂贵诉讼或因分区法变更而被取消的公司来说，等待也可能值得。
不过，请记住，Oklo的商业计划和技术都具有高度投机性，使其股票目前来说是一个非常冒险的买入选择，具有充足的波动性和事情出错的潜力。只有最具有风险承受能力的投资者才应该考虑现在购买Oklo的股票。但如果Oklo能够创造出对核能供电的数据中心的需求，那么它的股票可能会成为一个巨大的长期赢家。
在您购买Oklo的股票之前，请考虑以下几点：
Motley Fool Stock Advisor的分析师团队刚刚确定了他们认为投资者现在应该购买的10支最佳股票……而Oklo不是其中之一。这些股票可能会在未来几年产生巨大的回报。
请考虑Netflix在2004年12月17日被列入此名单时的情况……如果您当时投资了1,000美元，您将拥有463,900美元！或者，当英伟达在2005年4月15日被列入此名单时……如果您当时投资了1,000美元，您将拥有1,294,401美元！
值得注意的是，Stock Advisor的总平均回报率为978%——与标准普尔500指数的211%相比，这是一个市场表现优于市场的回报。不要错过最新的前10名名单，该名单可使用Stock Advisor，并加入由个体投资者为个体投资者建立的投资社区。
Stock Advisor的回报截至2026年5月30日。
John Bromels持有亚马逊和Oklo的股份。富达共同基金持有亚马逊和Vertiv的股份。富达共同基金的披露政策。
本文中的观点和意见仅代表作者的观点和意见，不一定代表纳斯达克公司的观点。
四大领先AI模型讨论这篇文章
"Regulatory and construction timelines for nuclear make Oklo's co-location solution unlikely to scale before opposition and competition erode its window."
The article frames Oklo as a potential fix for data-center opposition via co-located SMRs, citing better nuclear polling and efficiency gains with Vertiv. Yet Oklo remains pre-revenue with its 75 MW Aurora prototype still two years from operation and zero disclosed co-location designs or customer contracts. Nuclear siting, NRC licensing, and waste rules impose multi-year delays that local zoning fights rarely match, while excess power sales to grids reintroduce the very utility-cost concerns driving 71% opposition. Execution risk dwarfs the speculative upside until first power is demonstrated.
Fast-tracked DOE-backed projects and bipartisan nuclear support could compress timelines enough for Oklo to lock in hyperscaler deals before traditional data-center builds resume.
"Oklo's thesis requires solving three unproven problems simultaneously (SMR commercialization, Vertiv integration, local acceptance) while the article treats the third as already solved by nuclear's polling advantage."
The article conflates two separate problems: local opposition to data centers (real, documented) and a speculative solution (Oklo SMRs). The 71% opposition figure is genuine, but the leap to 'nuclear solves this' ignores critical gaps. Oklo hasn't built a single commercial reactor. Its Aurora prototype is 75 MW—enough for one modest data center, not the gigawatt-scale buildout AI requires. The Vertiv partnership is announced but unproven. Most critically: rural communities opposing data centers may oppose nuclear plants equally. The article assumes zoning resistance is purely about grid strain, not land use, noise, or perceived risk. Until Oklo demonstrates one operational facility, this is venture-stage speculation dressed as infrastructure inevitability.
If Oklo executes and proves SMRs can be sited faster than grid-connected data centers, the addressable market is enormous—and early investors capture asymmetric upside before competitors enter.
"Oklo's thesis relies on a regulatory and technical timeline that is vastly more complex than the article suggests, making it a high-risk gamble rather than a viable infrastructure solution."
The narrative that Oklo (OKLO) solves the 'NIMBY' (Not In My Backyard) problem for data centers via SMRs is premature and fundamentally ignores the regulatory reality. While the partnership with Vertiv (VRT) is a logical synergy for cooling, the article glosses over the immense licensing hurdles of the Nuclear Regulatory Commission. Even if the tech works, the 'co-location' strategy doesn't bypass local zoning or public safety concerns; it likely intensifies them. Investors are pricing in a commercial breakthrough that is years away, ignoring that data center operators prioritize uptime and low-cost power, not experimental modular reactors with unproven supply chains. The $64 billion in stalled projects won't be saved by a prototype that hasn't cleared its first deployment.
If Oklo successfully achieves its first operational deployment, the regulatory 'first-mover' advantage could create a massive moat that renders traditional grid-dependent data centers obsolete.
"The core risk is regulatory and scalability: without a credible, financed proof-of-concept, Oklo's stock is a high-variance bet rather than a reliable path to transforming AI data-center infrastructure."
Oklo and Vertiv tout a co-located 75 MW SMR data-center concept aimed at reducing cost, water use, and siting friction. While appealing in theory, the piece glosses over critical hurdles: (1) regulatory risk and long NRC licensing timelines for SMRs; (2) scale economics—75 MW per site remains small for hyperscale AI workloads needing multi-hundred MW; (3) operational risks and public acceptance, including heat, water, waste, and decommissioning concerns. Until a proven, financed deployment proves cost, safety, and grid integration, the idea stays speculative and likely limited to a handful of pilot sites rather than a national wave.
Even if a pilot passes, licensing, insurance, and capital costs will drag timelines and returns; local opposition could re-emerge around nuclear contamination fears and water usage, undermining the 'easy path' narrative.
"Federal preemption may ease siting but fuel supply and waste logistics remain unaddressed execution risks for Oklo."
Gemini's focus on NRC licensing and local zoning overlooks federal preemption powers that can override community opposition for nuclear projects, unlike standard data-center builds. This creates a potential timeline advantage the article underplays. Yet the unmentioned risk is securing reliable fuel supply chains and waste disposal for scaled SMR deployments, which could create bottlenecks even after initial approvals, amplifying Oklo's execution challenges beyond pilots.
"Federal preemption doesn't eliminate state/judicial delays, and Oklo's unit size may be architecturally incompatible with hyperscaler redundancy mandates."
Grok's federal preemption argument is theoretically sound but historically overstated. Nuclear projects still face state-level challenges (California, New York) and judicial review that can delay years. More critically: nobody's addressed that hyperscalers themselves may reject Oklo's 75 MW units due to redundancy requirements—data centers typically demand N+1 or N+2 backup capacity. A single SMR failure isn't an option. That architectural mismatch could kill the thesis before NRC licensing even matters.
"High O&M costs for distributed SMRs will likely render them economically uncompetitive against grid-tied power, regardless of redundancy or zoning advantages."
Claude is right about the N+1 redundancy, but misses the capital efficiency angle. Hyperscalers don't need a single SMR to power a whole campus; they need them as distributed micro-grid baseload to avoid massive transmission upgrades. The real risk isn't just the NRC, but the 'stranding' of capital. If these reactors require dedicated security and specialized staff, the O&M costs per kilowatt-hour will likely dwarf natural gas or grid-tied renewables, making the business model DOA.
"Federal preemption alone won’t reliably accelerate Oklo’s path; fuel/waste logistics and SMR vendor lead times are the real, underappreciated bottlenecks."
While Grok argues preemption could accelerate approvals, I think that optimism ignores the practical gating factors that outlive zoning fights. Federal preemption can help, but state environmental reviews, tribal rights, and environmental impact statements still loom and can add years. More material: securing reliable SMR fuel supply and long-term waste management—and vendor lead times for modular reactors—could strangle scaling even after NRC licensing. The near-term path remains a multi-year, capital-intensive risk.
The panel is bearish on Oklo's co-located SMR data-center concept, citing regulatory hurdles, scale economics, operational risks, and potential architectural mismatches with hyperscalers' redundancy requirements.
Potential timeline advantage through federal preemption powers for nuclear projects.
Securing reliable fuel supply chains and waste disposal for scaled SMR deployments, as well as potential rejection by hyperscalers due to redundancy requirements.