Last week, Chinese tech stocks went through a mini rollercoaster.

The Wall Street Journal first reported that regulators were close to completing their cybersecurity investigations of Didi, Full Trust Alliance, and Kanzhun, sending stock prices up. Then Bloomberg reported that the China Securities Regulatory Commission (CSRC, China’s SEC) began the processing of reviving the highly-anticipated Ant Group IPO, sending stock prices up some more. All this enthusiasm was soon damped when CNBC reported that both Ant and CSRC denied that a new IPO is being discussed and worked on.

The whiplash of these headlines tend to cause further confusion of an already confusing situation, but it does seem to me that some kind of “end” is on the horizon. While most media narratives like to frame all this as a “crackdown”, another framing I find more clarifying is China removing its “startup debt”, which I first introduced in a previous post.

In short, every organization – a company, a bureaucracy, a country – needs to remove some amount of “debt” periodically in order to grow and evolve sustainably. If you analogize the Chinese economy to being a single company, its startup journey began in 1978 with Deng Xiaoping’s “opening and reform”. Since then, this “company” grew at a torrid pace, but this 40-plus-year-old startup has never made a concerted effort to clean up all its “startup debt” until 2020. If Jack Ma did not give that speech, China may have delayed this unavoidable task further.

(You can read our translation of Jack Ma’s infamous Bund Finance Summit speech HERE (premium content))

In other words, whether you call it a “crackdown”, “reform”, “removing startup debt” (my preferred framing), or something else, everything that has happened in the last two years is long overdue. Given all the news last week, it is worth extending this “startup debt” framing further and think about if/when the Ant IPO is back on track, what does it really mean.

The “Startup Debt” Experience

For readers who’ve worked in a tech startup before, the notion of “startup debt” should feel familiar. But for readers who’ve worked primarily in government, policy, academia, or even investment firms that don’t get involved in a company's operations, this concept can feel foreign, so I’ll do a quick explainer.

Every startup accumulates some amount of “debt” at various stages of its creation and growth. The original engineering team may have chosen the wrong stack or programming language to build the product. The company might’ve selected the wrong HR and benefits provider that did not offer good enough health benefits to retain employees over time. The founding CEO may have hired the wrong early exec to build its go-to-market team. These are all real examples of “startup debt”.

Accumulating “startup debt” is unavoidable and not necessarily a bad thing – it simply reflects the set of tradeoffs and priorities that a company’s leader has to contend with at a given point in time. If you build the perfect tech stack, but have no user or customer, this “over building” can waste a lot of money and kill the company. If your product found traction and took off like a rocketship, but your backend architecture does not evolve fast enough to scale, that can also kill the company. It’s not easy to strike the perfect balance between driving growth or removing debt. And it usually takes a tech startup 8-10 years to reach a stage of maturity where “removing startup debt” is warranted.

For a company, these “debts” often come in the form of technical architecture, organizational structure, and business processes.

For a country, these “debts” often come in the form of regulatory clarity, institutional policymaking, and governmental capacity.

Every emerging economy is essentially an early stage startup at different stages of growth and maturity, where the question of “when should we start cleaning up our startup debt” always looms on the horizon.

It’s worth noting that this question does not only apply to young companies or emerging economies. Mature companies and countries need to answer it too or risk decline. The US Senate’s current effort to update America’s woefully outdated antitrust regulations and introduce a cryptocurrency bill are both examples of a mature economy doing so.

Famous Tech Debt Examples

Before we talk about the Ant IPO and how to think about China’s overall “startup debt” removal progress, let’s look at a few well-known tech debt removal examples, in brief, to put in context just how complicated these undertakings are, even in the well-defined, well-scoped context of a single company. We will look at three examples that span both the US and China tech sector: Facebook, Dropbox, and Alibaba.

Facebook (From “Break Things” to “Stable Infra”):

The most high-profiled startup tech debt removal is Facebook (or Meta)’s transition, from the controversial “Move Fast and Break Things” ethos, to the decidedly more boring “Move Fast with Stable Infra”. This transition began in 2014, exactly 10 years after Facebook's founding.

It’s safe to say that the progress of this transition is still on-going and nowhere near completion. For example, Instagram suffered downtime as recently as November 2021. As the company evolves into a full-blown Metaverse company, its infrastructure will have to continue evolving to support this new paradigm. Facebook was famously a MySQL shop, a popular open source relational database that is fast and reliable with small-size, transactional data. This technology won’t work with social networking in a VR world, where everything is real-time interactive and image-heavy.

In summary:

  • Facebook’s time from founding to “tech debt removal” transition? 10 Years
  • Time to completion? TBD (7 years and counting)

Dropbox (From AWS to Private Cloud):

The cloud storage company, Dropbox, once a darling of Silicon Valley, had one of the most epic “tech debt” removal stories. A few years prior to its IPO in 2018, it transitioned from using AWS to building its own private cloud with custom servers, all the while maintaining its service with no downtime. This transition was necessary both as a massive cost-saving measure and to reduce its strategic dependency on AWS. The whole story was dramatically told in this Wired magazine article. (For more of my thoughts on Dropbox, see previous articles “Should Dropbox Enter the Cloud Business?” Part 1 and Part 2)

In summary:

  • Dropbox’s time from founding to “tech debt removal” transition? 6 Years
  • Time to completion? 2.5 years
Dropbox custom-designed servers. Photo by: Christie Hemm Klok/WIRED

Alibaba (From IOE to AliCloud):

When Alibaba was founded in 1999, the entire Chinese tech sector was dependent on American infrastructure technology vendors, both hardware and software. The so-called “IOE” stack, as in IBM, Oracle, EMC, makes up the standard stack of server, database, and storage that every tech company started on, including Alibaba, Tencent, and Baidu.

As Taobao and AliPay took off, Alibaba’s infrastructure cost also skyrocketed, presenting a similar challenge that Dropbox was wrestling with vis-a-vis AWS. Thus, Alibaba began removing its “tech debt” in 2009 by starting the “remove IOE” initiative, documented in this Zhihu thread. The slide below was part of an internal Taobao presentation, detailing the transition goals:

  • From IBM to commodity PC server
  • From Oracle to MySQL (yes, the same open source MySQL Facebook uses which Oracle later bought via its Sun Microsystems acquisition in 2010)
  • From EMC to mid to lower-grade storage unit

This transition was completed in 2013 and paved the technical foundation and expertise needed to build AliCloud. It's also safe to say that without this “removing IOE” initiative, Ant Group would not be where it is today

In summary:

  • Alibaba’s time from founding to “tech debt removal” transition? 10 Years
  • Time to completion? 4 years

So how should we use these examples by companies to think about the Ant IPO and China’s country-wide process of “removing startup debt” at large?

Ant IPO: Timing and Significance

There is a good chance that the Ant IPO will resume in earnest some time in 2022. If so, that would put the time between this IPO getting pulled down to then revived at two years.

Two years is not a long time, in the context of “removing startup debt”. Among the three examples I cited, even the most well-defined, well-scoped project, namely Dropbox’s, took 2.5 years!

The thing with “removing startup debt” is while the cleaning is going on, you can’t stop operating, stop growing, or stop experimenting with the next new thing. One big, new thing that the Ant IPO is supposed to help answer is how will a high-profiled dual-listing on the Hong Kong and Shanghai stock exchange play out, where foreign investment interest will be extremely high? Will it be a free for all, or will there be restrictions and ownership limits placed on foreign capital?

What the Ant IPO can answer is just one dimension of a myriad of “startup debt” that China is removing, in order to grow more sustainably.

As I mentioned in my previous post, China’s “startup debt” is multifaceted and accumulated over a multi-decade period, which is partially why the so-called “crackdown” from two years ago felt so intense. Some of this “startup debt” is literally just debt (like Evergrande). Some of this debt is technical (like the new data governance, data privacy, and cybersecurity regulations). Some of this debt is regulatory (like antitrust, fintech reform, foreign and domestic IPO, and capital market access). Some of this debt is cultural (like outlawing for-profit after-school tutoring and limiting minors’ videogame playing time).

While a few of these areas are making progress and inching closer to clarity, other realms are still being cleaned out and will take a long time to complete. Coupling this with the disruptive execution of the dynamic zero-Covid policy, completely removing China’s “startup debt” will be a decade-long process, not a two year sprint.

Thus, a 2022 Ant IPO should not be seen as the beginning of the end, but the end of the beginning.

蚂蚁上市与中国的 "创业债" 问题



这些头条往往会使本已混乱的局面更加混乱,但在我看来,某种 "终点" 似乎即将到来。虽然大多数媒体喜欢把这一切归结为 "镇压",但我觉得用另一个框架来看事情更清晰,那就是中国重在清除其 "创业债",我在之前的一篇文章中首次介绍了这个框架。

简而言之,每个组织, 不管是个公司、政府机构、还是国家,都需要定期清除一些 "债",才能可持续增长和发展。如果你把整个中国经济比喻为一家公司,它的创业历程始于1978年启动的 "改革开放"。从那时起,这家 "公司" 以迅猛的速度增长,但这家有40多年历史的“创业公司”也从未作出努力来清理其所有的 "创业债",直到2020年。如果马云没有发表那次在外滩金融峰会的演讲,清理“创业债”这堆事可能还会继续推迟。

换句话说,无论你把事情称之为 "打压"、"改革"、"清除创业债"(我喜欢的框架),还是其他什么,过去两年发生的一切都早该发生了。鉴于上周的所有头条,值得进一步延伸这个 "创业债" 的思维框架,并思考如果那天蚂蚁再次IPO,它真正意味着什么。

思考体验 “创业债”

对于那些曾经在科技创业公司工作过的读者来说,"创业债" 这个概念应该很熟悉。但对于那些主要在政府、政策、学术界,甚至是不参与公司运营的投资机构工作的读者来说,这个概念可能有些陌生,所以我来做一个简短的解释。

每家创业公司在其创建和成长的不同阶段都会积累一定量的 "债"。公司可能选错了人力资源和福利供应商,因此没有提供足够好的健康福利来长期保留员工;创始CEO可能聘错了来搭建市场团队的早期高管。这些都是 "创业债" 的真实例子。

积累 "创业债" 是不可避免的,也不一定是件坏事 – 它只是反映了一个公司的领导团队在某一特定时间点所必须面对的一系列权衡和对各种项目的优先级选择。如果你搭建了完美的技术堆栈,但没有用户或客户,这种 "过度搭建" 可能会浪费大量资金并扼杀公司。如果你的产品找到了牵引力,并像火箭一样飞速增长,但后端架构没有足够快的发展来扩容,这也会扼杀公司。要在推动增长或清除债之间取得完美的平衡极为不易。一家科技创业公司通常需要8-10年的时间才能达到一个足够成熟的阶段,值得去花时间清除自己的“创业债”。

对于一家公司来说,这些 "债" 通常是以技术架构、组织结构和业务流程的形式出现。

对于一个国家来说,这些 "债" 往往以监管清晰度、机构决策和政府能力的形式出现。

每个新兴经济体本质上都是一个处于不同发展和成熟阶段的早期创业公司,"应该何时开始清理我国的创业债" 这个问题会一直徘徊。



在讨论蚂蚁上市以及如何看待中国整体的 "创业债" 清除过程之前,让我们简单看看几个著名的“技术债” 清除案例,以了解这种事情有多复杂,即使是在定义明确、范围清晰的一家公司内部的背景下。我们将看一下跨越美国和中国科技领域的三个例子:Facebook、Dropbox和阿里巴巴。

Facebook(从“Break Things” 到 “Stable Infra”)

最知名的科技公司“技术债”就是Facebook(或Meta)的转型,从有争议的 "Move Fast and Break Things" 调调转型到明显更无聊的 "Move Fast with Stable Infra"。这一转型始于2014年,即Facebook成立整整10年之后。



  • Facebook从创立到 "去除技术债" 的过渡时间?10年
  • 到完成的时间?待定(7年,还没做完)


云存储公司Dropbox曾经是硅谷的宠儿,有一段最史诗般的 "技术债" 清除故事。在2018年上市前几年,它从使用AWS过渡到用定制的服务器建立自己的私有云,同时维持其服务,没有停机。这种过渡是必要的,既是一种大规模的成本节约措施,也是为了减少其对AWS的战略依赖性。整个故事在这篇《Wired》杂志的文章中像电影剧本似的描述了。(关于我对Dropbox的更多分析,请参见以前的文章 《Dropbox应该步入云业务吗?》 第一篇第二篇。)


  • Dropbox从创立到 "去除技术债" 的过渡时间?6年
  • 到完成的时间?2.5年


当阿里巴巴在1999年成立时,整个中国科技领域都依赖于美国的基础设施供应商,从硬件到软件。所谓的 "IOE" 堆栈,IBM、Oracle、EMC,是每家科技公司起步时的毫无选择的服务器、数据库和存储的标准堆栈,包括阿里、腾讯和百度。

随着淘宝和支付宝的腾飞,阿里的基础设施成本也急剧上升,类似于Dropbox与AWS之间的关系。因此,阿里在2009年开始清除其 "技术债",启动了 "去除IOE" 计划,在这条知乎帖子有详细描述。下面的slide是淘宝内部讨论去除IOE的详细介绍和转型目标。

这一转型在2013年完成,为建立阿里云也铺垫了技术基础和专业知识。也可以说,如果没有 "去除IOE",也不会有蚂蚁今天的成就。


  • 阿里巴巴从创立到 "去除技术债" 的过渡时间?10年
  • 到完成的时间?4年

了解这些“技术债”案例后,我们应该如何思考蚂蚁的IPO和中国整体在清除 “创业债” 的进程呢?



就 "清除创业债" 而言,两年时间并不长!在我列举的三个例子中,即使是定义最明确、目标最清晰的项目,即Dropbox,也花了2.5年时间!


蚂蚁上市能够回答的仅仅是中国正在清除的无数种 "创业债" 的其中一个层面。

正如我在前一篇文章中提到的,中国的 "创业债" 是多方面的,而且已经积累了几十年,这也是为什么两年前开始的所谓 "打压" 感觉如此强烈的部分原因。这些“债”中的一部分真的就是债(如恒大),有些是“技术债”(如新出炉的数据治理、数据隐私和各种网络安全规则),有些是“监管债”(如反垄断、金融科技行业改革、国内以及海外IPO监管和资本市场的介入),有些则是“文化债”(如取缔营利性的课外辅导和限制未成年人玩游戏的时间)。

虽然某些领域已经取得进展,未来逐渐开始清晰,但许多其他领域仍在“清理中”,还需要很长一段时间才能完成。再加上“动态清零”政策的影响,完全清除掉中国的 "创业债" 将是一个十年的过程,而不是两年冲刺就能解决的。

因此,2022年蚂蚁如果上市不应该被视为是beginning of the end,而是end of the beginning。