字号·· | 护眼
亚洲新闻台

微软故障表明,大型科技公司不能被当作“大到不能倒”Microsoft outage shows that big tech can’t be treated as too big to fail

点「原文对照」整页切到原文,或双击某段只看那段的原文。

2026年11月13日拍摄的插图中可见微软标志和“全球宕机”字样。路透社/达多·鲁维奇/插图Microsoft logo and words "Global Outage" are seen in this illustration taken on November 13, 2026. REUTERS/Dado Ruvic/Illustration

上周末,微软遭遇故障,数字世界的大部分随之陷入停摆。此次中断瘫痰了 Outlook、Teams 和 Copilot 等关键微软服务,全球有超过 4.3 亿人使用这些服务。随着电子邮件和内部通讯冻结,政府和企业的运作也随之陷入瘫痪,引发的经济停摆可能造成数十亿美元损失。

hit Microsoft late last week, much of the digital world ground to a halt. The outage wiped out key Microsoft services including Outlook, Teams and Copilot, which are used by more than 430 million people around the world. As email and internal communications froze, the operations of governments and businesses went dark alongside them, triggering an economic paralysis that could cost billions of dollars.

在全球经济承受这场停摆的主要冲击之际,微软面临严重的声誉受损,以及因未能提供服务而可能高达每天 8 亿美元的合同违约金。

While the global economy absorbs the brunt of this paralysis, Microsoft faces both a severe reputational hit and contractual penalties that could reach up to US$800 million per day for services they failed to deliver.

如此灾难性的中断凸显了现代数字基础设施的脆弱。它依赖于少数几家大型科技公司,这些公司实际上已变得“大到不能倒”。我们究竟是如何陷入这种局面的?有没有出路?

Such a catastrophic outage shows just how precarious our modern digital infrastructure is. It depends on a handful of big tech companies, which have essentially become too big to fail. So how did we end up in this situation? And is there a way out?

数字依赖 微软此次故障的初始起因极其不起眼。数据中心通常在全球各地的系统间共享用户需求负载。然而,上周一个微小的配置变更导致公司失去了对部分计算机的控制。

DIGITAL DEPENDENCY The initial source of Microsoft’s failure was incredibly innocuous. Data centres usually share the load of user demands across systems all over the world. However, last week, a small configuration change caused the company to lose control of some of their computers.

无法分担负载,这些计算机将用户请求转嫁到其他系统,使其不堪重负并引发更多中断。这种模式反复发生,直到公司几乎完全失去对其网络的控制。

Unable to share the load, these computers dropped user requests onto other systems, overwhelming them and producing more outages. This pattern repeated over and over again until the company lost almost complete control of their network.

更糟的是,断连的用户设备——手机、电子邮件客户端和业务系统——全部自动尝试同时重新连接,再次导致微软系统全面崩溃。

Worse, disconnected user devices – phones, email clients and business systems – all automatically tried to reconnect at once, crashing Microsoft’s systems all over again.

历史上,技术故障可能只影响单一公司。但微软的中断之所以能波及数百万人,归根结底是一场悄然发生、已持续 20 多年的商业革命。

Historically, a tech outage might affect a single company. But the reason Microsoft’s outage could impact millions of people comes down to a business revolution that silently took place over 20 years ago.

软件曾是一次性以预付成本出售给公司的产品。在这种模式下,购买方拥有很大权力——只有当更新值得花费时,才会购买新版本。

Software was once something that was sold to companies once for an upfront cost. In this model, purchasers held great power – only buying new versions if the update justified the expense.

然而,这些前期支出也吓跑了消费者。电信、健身房和房地产租赁行业为科技行业提供了灵感,科技行业转向出租其产品的使用权,创造了如今被称为“软件即服务”的模式。

Yet, these upfront expenses also scared consumers off. Drawing inspiration from telecommunications, gyms and real estate leasing, the tech industry transitioned to renting access to their products, creating what is now known as “software as a service”.

理论上,每个人都从这种模式中获益。软件供应商能够获得更可靠的现金流,而客户则可以改为定期支付较小的款项,而不是一次性支付高昂的前期费用。

In theory, everyone profited from this model. Software vendors could bank more reliable cash flow, and customers could transition to regular smaller payments, rather than stumping up for large upfront costs.

最终,这种变化将权力天平向科技公司倾斜。消费者不再购买所需的功能,而是将运营王国的钥匙交给了“软件即服务”提供商,并在此过程中将自己锁定在极其昂贵的长期合同中。

Ultimately, this change shifted the balance of power towards tech companies. Instead of purchasing features they needed, consumers handed the keys to their operational kingdoms to the “software as service” providers, and locked themselves into highly expensive long-term contracts in the process.

太大而不能倒闭 2008年全球金融危机向世界介绍了“太大而不能倒闭”的概念。危机爆发时,公众不得不收拾那些因过于系统重要性而不被允许倒闭的银行的残局。

TOO BIG TO FAIL The global financial crisis in 2008 introduced the world to the idea of too big to fail. When it hit, the public was left to pick up the pieces of banks that had become too systemically important to be allowed to fail.

不幸的是,看来我们并没有真正吸取这些教训。相反,过去几年我们一直在经历科技行业的一个慢动作版本。集中化、整合以及不受约束的垄断,使我们的数字生活、企业和政府处于极其危险的境地。

Unfortunately, it appears we haven’t really learned those lessons. Instead, we’ve spent the last few years living through a slow-motion version of that with the tech industry. Centralisation, consolidation and unchecked monopolies have left our digital lives, our businesses and our governments in an incredibly precarious place.

在某个层面上,互联网上几乎所有东西都经过谷歌、微软、Meta、亚马逊云服务或CrowdStrike——这些公司近年来都曾发生过重大宕机。

e level or another, almost everything on the internet goes through Google, Microsoft, Meta, Amazon Web Services or CrowdStrike – all of which have had major outages over recent years.

这些科技巨头凭借提供统一、看似不可战胜的基础设施赢得了我们的信任。但集中化带来的便利也埋下了陷阱。

These big tech giants earned our trust by providing unified, seemingly invincible infrastructure. But the convenience that comes with centralisation holds a trap.

与2008年的银行业不同,当这些科技巨头出现故障时,他们不需要请求政府救助。相反,他们的核心地位迫使整个数字经济承担其宕机造成的成本。发展数字主权摆脱对少数科技巨头在国家基础设施、企业及公民数据方面的依赖,推动了所谓“数字主权”的呼声。

Unlike the banking sector in 2008, when these tech titans fail, they don’t need to ask governments to bail them out. Instead, their centrality forces the entire digital economy to bear the costs of their outage. DEVELOPING DIGITAL SOVEREIGNTY Escaping the reliance on a few tech giants for our national infrastructure, businesses and citizens’ data has spurned the push for so-called digital sovereignty.

欧洲在这一领域走在前列。通过积极资助去中心化云基础设施和本土软件,各国政府正在确保美国科技故障不会导致当地医院瘫痪。

Europe is leading the way on this. By actively funding decentralised cloud infrastructure and homegrown software, governments are ensuring that an American tech failure cannot knock out a local hospital.

世界必须采取类似的蓝图。澳大利亚尤其具备把握数字主权带来机遇的优势。我们已拥有原始的工程人才。Atlassian、Apromore、Airwallex 和 Canva 等公司已证明,我们能够构建世界级的数字工具。

The world must adopt a similar blueprint. Australia, in particular, is well placed to grasp the opportunities that digital sovereignty brings. We already have the raw engineering talent. Companies such as Atlassian, Apromore, Airwallex and Canva have proven we can build world-class digital tools.

如果政府在本地科技生态系统上进行战略性投资,不仅是在购买“数字保险”,还将推动亟需的市场竞争,稀释大型科技垄断的权力,分散系统性风险——确保单一故障点永远无法让经济跪倒在地。

If governments strategically invested in local tech ecosystems, they wouldn’t just be buying digital insurance. They would drive much-needed market competition, dilute the power of big-tech monopolies and spread out systemic risks – ensuring that a single point of failure could never bring the economy to its knees.

这并非说我们应切断与大型科技平台的一切联系,这些平台确实并将继续提供有用的服务。但如果全球经济继续依赖少数“大而不能倒”的数字垄断,当我们的屏幕不可避免地再次变黑时,我们只能怪自己。

This is not to say we should divest ourselves of any connection to big-tech platforms, which do and will continue to provide useful services. But if our entire global economy continues to rely on a handful of too-big-to-fail digital monopolies, we will have no one but ourselves to blame when our screens inevitably go black again.

Andrew Cullen 是墨尔本大学计算与信息系统学院客座研究员。本评论首发于 The Conversation。

Andrew Cullen is Guest Research Fellow at the School of Computing and Information Systems, University of Melbourne. This commentary first appeared on The Conversation.

阅读媒体原文 ↗