Open Source in China: The Trends

This is the third and last part of my multi-part series exploring “open source in China”. If you missed the two earlier posts, I’d strongly recommend you read them first to build some context. (Part I, Part II)

In this post, I’ll wrap up the exploration with three macro-trends that I believe will drive open source in China for the next five years: the rise of cloud computing platforms, new investments from all corners of the tech community, and geopolitical necessities in the forms of self-reliance and influence abroad.

Rise of the Cloud

Cloud platforms (e.g. AWS, Azure, etc.) have fundamentally changed how open source technologies are being distributed. Frictionless distribution used to be one of the key advantages of open source over its proprietary competitors. That’s no longer the case. You can almost think of the cloud as an app store for open source. Having more open source technologies available on a cloud platform makes that platform more valuable too. The value creation dynamic goes both ways.

With the cloud growing rapidly in China, the same distribution dynamic will take off in a big way. In Q4 of 2019, China has become the second largest cloud market in terms of spend. This growth will lead to more open source projects being created and distributed in both a cloud-native way. Open source creators and developers in China will have “cloud” at the top of their mind.

This distribution vector will not be limited to inside China. Out of the big Chinese tech companies that operate public cloud platforms, most of them have data centers abroad. Alibaba Cloud is the leader of the pack, as I’ve analyzed in a previous article: “Where Are The Data Centers: AWS, Alibaba Cloud, Azure, GCP”. Tencent Cloud also has data centers and colocation facilities around the world. Alibaba is stepping up its cloud investment with an eye-popping $28.2 billion USD investment over the next three years, which I’ve argued in a previous post will lead to data center expansions in Southeast Asia and Latin America plus additional investments inside China. Because open source technologies are borderless by default, ones first created in China are or will be on all the cloud platforms run by American tech giants too.

Of course, the actual distribution will take a few years to play out. It is no trivial engineering work to make an open source project operate well on a cloud, let alone multiple clouds. And the rollout from one cloud Region to another within a given cloud platform also takes time, and depends on business demand and technical resources. That being said, all cloud platforms are converging towards Kuberentes, an open source container orchestration technology, which will drive down the cost of integrating open source projects across different clouds significantly.

New Investments Piling In

As open source grows in China, naturally, more investments of all types will be piling in. The targets of these investments won’t be limited to new and existing commercial open source startups, but also conferences, community meetups, hackathons, and other (stereo)typical open source developer activities. These offline events play an essential role in the ecosystem, and the best way to earn some goodwill among developers is to buy the pizza, beers, and maybe a side of spicy crawfish. I see three sources of investments, each with its own implications:

Venture Capital

As I’ve noted in Part I of this exploration, Chinese VCs are shifting their focus towards B2B enterprise startups, as consumer tech opportunities dry up. Commercial open source startups make up a big portion of enterprise tech, because almost all of them innovate in different layers of the infrastructure stack. Two potential problems I see with the influx of VC dollars.

One problem is the lack of notable VC-backed successes in the enterprise space thus far. UCloud’s listing earlier this year on China’s new domestic exchange, STAR, was a modest outcome at best. Kingsoft, the OG of enterprise software in China, was founded in 1988, long before venture capital was a thing. To be fair, many Chinese VCs study the US market intensely to become sharper investors and adapt learnings to their domestic market. But vicarious, second-hand learnings can only take you so far. Until a winner comes along, the effects of piling on more investment will just be mo’ money, mo’ problems.

The other problem is that the new VC entrants to the open source ecosystem all made their money betting on hyper-growth, consumer tech companies. Building B2B enterprise businesses, open source or not, require both more patience and a completely different way of looking at growth, business models, and company building. Further complicating the matter is that certain metrics of open source projects can look deceivingly “consumer-ish”. GitHub stars and the number of forks, clones, and commits are some of the usual culprits. Without a deep understanding of what makes open source successful and what makes commercializing open source work (two separate questions), VCs who are used to the consumer tech way of building startups will lead companies astray.

Big Tech

We may see more aggressive strategic investments from Chinese tech giants into commercial open source startups. Thus far, the only domestic example is Baidu’s modest investment in Gitee. Outside of China, Alibaba has led an investment in MariaDB, the open source fork of MySQL.

While China’s consumer tech ecosystem has been mostly carved out by the tech giants via strategic or controlling investments, most notably Alibaba and Tencent, the same has not occurred in the emerging enterprise space. That’s partly due to the lack of maturity in the space overall. Another reason is resistance from founders, especially of commercial open source startups, who are concerned about the perception (or reality) of platform lock-in, which is smart thinking on their part. As open source technology becomes more reliant on the cloud for distribution, taking a strategic investment from a big tech company that likely also operates its own cloud would hurt the technology’s long-term adoption and commercial potential.

Other Strategic Players

Strategic investments of a different flavor could also come from large enterprises, like banks, insurances, telecoms, and manufacturing. Some possibilities are China Unicom, China Mobile, China Telecom, Ping An Technology (part of Ping An Insurance), China Merchant Bank, and SAIC Motor, all of which are dabbling in open source with either technology adoption or memberships with the Linux Foundation or its subsidiary, the Cloud Native Computing Foundation.

For commercial open source startups, investments from these players may present interesting growth opportunities and technical partnerships, while breaking into some lucrative industries with no platform lock-in concerns. However, there are tradeoffs to every relationship. Because these large enterprises are all heavily regulated by the Chinese government, if and when an investment like this does happen, how that will affect the open source technology’s adoption in and outside of China and its international business potential is an open question.

Geopolitical Necessities

The interconnections between open source and a country’s geopolitical necessities are as complex as complexity comes. I won’t be able to dive into each rabbit hole in this post, but there are two trends I think people should pay attention to: RISC-V and China Standards 2035.

RISC-V

RISC-V is an open standard Instruction Set Architecture (ISA) for hardware, first originated in the academic confines of UC Berkeley and available via open source licenses. This means anyone can run, change, copy, and distribute it, in accordance with the four freedoms that underpins the worldview of open source. In layman's terms, an ISA is basically the set of instructions needed to build the heart of a computer, i.e. the CPU.

As an open source ISA, RISC-V has massive potential to disrupt and unlodge the incumbency in the chip design industry dominated by the old guards, like Intel and ARM. After the sanctions against Huawei and ZTE during the height of the U.S.-China trade war, it’s no news that China is desperately trying to be less dependent on imported semiconductors, all of which are based on proprietary designs. By embracing and supporting the RISC-V ecosystem and being involved in the community stewarded by the RISC-V Foundation, China may very well accelerate its pace towards self-reliance. China is by no means the only country seeking such self-reliance; India is another major example.

At the moment, few RISC-V silicons are actually being manufactured by the major chip foundries, like the Taiwan Semiconductor Manufacturing Company (TSMC). Most of the big companies embracing RISC-V are using it as a bargaining chip to negotiate better deals with ARM, who still has a stranglehold on the market. RISC-V startups, like SiFive, mostly show traction and progress with design wins. I see Chinese tech companies being the main driver to catalyze RISC-V designs into production-grade silicons at scale in the next five years, due to both business imperatives and geopolitical necessities.

China Standards 2035

China Standards 2035 is a massive policy initiative from the central government to standardize all the specs of emerging technologies, from cloud computing and big data, to AI, VR, and 5G. Most of the standards will be applied domestically, but there is also indication that China would like to export some of those specs to influence how these technologies are standardized internationally. This proactive, some would call aggressive, position is causing some people heartburns. Some of the concerns from businesses and industry leaders are quite reasonable, other concerns from so-called “experts” with an agenda to push are rather shrill and baseless. The policy framework itself has only been in the works for two years. More details are supposed to be announced later this year. Until then, it’s too early to judge either way.

What does this have to do with open source? Most of the technologies being standardized are already dominated by open source solutions. Chinese companies will need to use those solutions just to stay innovative. In the process, more of China’s tech community will have to embrace the open source way of doing things, like transparent governance, open discussions with stakeholders and developers, and fair procedures for rulemaking. Absorbing these elements of open source will not only help China’s standardization efforts domestically, but also help the country become a responsible international shareholder with the integrity needed to be seen as a trustworthy player. Without that trust, any attempt by China to exert influence abroad will be caricatured as unsophisticated aggression by the “new rich”. That perception, justified or not, will not help any technology progress in the right direction.

(Please see Part I and Part II of this series exploring open source in China, if you missed them.)

If you like what you've read, please SUBSCRIBE to the Interconnected email list. New posts will be delivered to your inbox (twice per week). Follow and interact with me on: Twitter, LinkedIn.

中国的开源世界:未来趋势

这是我探索“中国的开源世界”系列的第三篇文章,也是最后一篇。如果您错过了前两篇文章,我建议您先读读它们构建一些backgrond context。(第一篇第二篇

本篇文章会分析三个主要的宏观趋势,我觉得会在未来五年内推动开源在中国的发展:云计算平台的崛起,来自科技界各个角落的新投资,和以自力更生和海外影响力形式表现出的地缘政治和国际关系的需求。

云的崛起

云平台(如AWS、Azure等)从根本上改变了开源技术的分发方式。无摩擦的分发和使用曾经是开源相对于其闭源竞争对手的主要优势之一。现在已经不是这样了。你几乎可以把云看作是开源的app store。在云平台上提供更多的开源技术也使该平台更有价值。这里的价值创造关系是双向的。

随着云计算在中国的迅速发展,中国的开源项目也会得到同样的分发优势。2019年第四季度,中国从消费角度来看已成为全球第二大云计算市场。这种增长将导致更多的开源项目以“云原生”的方式构建和分发。在中国,开源创造者和开发者会把“云”放在最重要的位置。

这种分发向量不限于中国境内。在有自己的公有云平台的中国科技巨头中,大多数都在国外设有数据中心。正如我前一阵子的一篇文章中 “AWS、阿里云、Azure、GCP的数据中心都在哪儿?” 所分析的,阿里云是领头羊。腾讯云在全球也有数据中心和托管设施。阿里正在加大对云的投资,在未来三年内将投2000亿人民币,这一数字令人瞠目结舌。我也曾经分析过,这笔投资的结果很有可能是向东南亚和拉丁美洲扩张,同时继续在中国境内加力。开源技术是无边境的,所以从中国开始的开源技术也已经或在不久将来都出现在所有美国科技巨头的云平台上

当然,具体发行一般需要几年时间才能完成。使一个开源项目在一个云上运行良好,更不用说在多个云上,并不是一件小工程。在某个云平台内,从一个云区域到另一个云区域的部署也需要时间,并且取决于商业业务需求和技术资源。目前一个趋势是所有云平台都在向Kuberentes(一种开源容器编排技术)靠拢,这将不小的降低跨越不同云平台部署集成开源项目的成本。

新投资大量涌入

随着开源在中国的快速发展,自然会吸引各类投资涌入。这些投资的目标将不仅限于商业开源初创企业,也会包括各种展会、社区活动、黑客马拉松和其他典型的开源开发者活动。这些线下活动在整个生态中起着至关重要的作用,如果想赢得开发者的好感,最直接的方式就是赞助pizza,啤酒,再加上些麻辣小龙虾。会有三种投资来源,每种都会对整个开源生态有不用意义的影响:

风险投资

正如我在这一系列探索中的第一篇文章中指出的,随着to C科技产品机会的枯竭,中国许多风投正将重心转向B2B企业服务初创。商业开源初创在企业技术服务中占很大一部分,因为几乎所有的开源企业都是在基础设施的不同层面上创新。在我看来,风险投资的涌入有两个潜在的问题。

一个问题是迄今为止得到风投的企业科技公司还缺乏显著的成功例子。UCloud今年早些时候在科创板上市,充其量只是一个OK的结果。金山软件是中国企业软件行业的老牌,但成立于1988年,那时候还没风投这回事儿呢。公平地说,许多中国风险投资人都在对美国市场进行深入研究,让自己的投资眼光更加敏锐,并将所学到的知识应用于国内市场。但是间接的、二手的知识毕竟有限。在一个大赢家出现之前,砸更多钱的结果很可能只是“钱多事儿更多”。

另一个问题是,新加入开源生态的风投许多以前都是投to C项目的。打造一个靠谱的B2B企业科技业务,无论技术是否开源,都需要更多的耐心和完全不同的方式看待增长、商业模式和公司建设。让问题进一步复杂化的是,开源项目的很多指标看似很像to C的产品,有误导性。GitHub的标星和forks、clones和commits的数量是常见的罪魁祸首。如果不深入了解开源成功的缘由和开源商业化的原因的话(两个截然不同的问题),那些习惯于投 to C初创的风投们将会把商业开源初创引向歧途。

巨头们

行业里可能会看到中国巨头们对商业开源初创企业进行更积极的战略投资。迄今为止,国内唯一的例子是百度对Gitee的小笔投资。在海外,阿里领投了MySQL的开源分支,MariaDB

虽然中国市场的to C地盘大多数都被巨头们的战略或控制性投资给瓜分了,但新兴的企业科技领域目前还没有出现同样的情况。部分原因是整个行业还不够成熟。另一个原因也许是founders的抵制,特别是商业开源初创。他们担心平台锁定的嫌疑(或现实),会影响整个开源项目的使用。这是个聪明有远见的想法。随着开源技术越来越依赖云来分发,如果接受一个自己也有云平台的巨头的战略投资,将损害技术长远的使用程度和商业潜力。

其他战略玩家

还有一种不同风味的战略投资,可能会来自与大型企业,如银行、保险、电信和制造业。中国联通、中国移动、中国电信、平安科技、招商银行和上汽集团都有可能做出这种投资。它们已经涉足开源领域,要么采用技术,要么加入Linux基金会或其子分支,云原生计算基金会(Cloud Native Computing Foundation),作为会员。

对于商业开源初创来说,这些大型企业的投资可能会带来一些独特的增长机会和技术层面的合作关系,打入几个钱多底子厚的大行业,还不用顾虑平台锁定问题。然而,每个关系都有权衡利弊。这些大型企业都受到中国政府的严格监管。如果这种投资真的发生,将会如何影响其开源技术在国内外的采用程度及其国际商业化的潜力,还是个未知数。

地缘政治需求


开源与一个国家的地缘政治需求的关联极为复杂,在这篇文章中,我也无法每个角度都深度讨论。从大方向看,我认为大家应该注意两个趋势:RISC-V和中国标准2035。

RISC-V

RISC-V是一种用于硬件设计的Instruction Set Architecture(ISA)开放标准,最初起源于加州大学伯克利分校的学术领域,可通过开源许可使用。这意味着任何人都可以运行、更改、复制和分发它,吻合开源世界观的四个自由。通俗地说,ISA基本上就是构建计算机核心(即CPU)所需的一组指令。

作为一个开源的ISA,RISC-V有着巨大的潜力来扰乱和解除由Intel和ARM等老牌企业为首的芯片设计行业。在中美贸易战最激烈时期对华为和中兴实施制裁之后,中国正在拼命减少对进口半导体的依赖。这已经不是什么新闻了。这些半导体都是基于专利设计的。拥抱和支持RISC-V生态,参与由RISC-V基金会主导的社区,很可能会加快中国走向自力更生的步伐。中国绝不是唯一一个寻求这种科技自力更生的国家;印度是另一个重要的例子。

目前,像台湾半导体(TSMC)这样的重头芯片生产商其实生产的RISC-V硅片很少。大多数拥抱RISC-V的大公司都把它作为与ARM谈判的筹码。ARM仍然是整个市场的领头羊。RISC-V初创公司,如SiFive,大多以“设计胜利”(deisgn win)来体现进展。我认为在未来五年,中国科技公司是推动RISC-V从设计到大批量硅片生产的主要动力,既出于商业需求,也是出于地缘政治需求。

中国标准2035

《中国标准2035》是中央政府出台的一项大规模政策举措,旨在规范新兴技术的所有规格,从云计算和大数据到人工智能、虚拟现实和5G。大部分标准将在国内应用,但也有迹象表明,中国希望出口其中一些规格,以影响这些技术在国际上的标准化。这种主动,有些人视为之咄咄逼人,已经开始让很多人担心。有些来自企业和行业领袖的担忧是合理的,而另一些来自所谓“专家”的担忧则是缺乏根据的。政策框架本身才筹备了两年。更多细节将在今年晚些时候公布。在那之前,任何判断都为时过早。

这和开源有什么关系呢?许多包括在《中国标准2035》需要标准化的技术已经被开源技术的解决方案所主导。中国企业需要采用这些开源方案来保持创新。在此过程中,更多的中国科技界人士将会吸收很多“开源做法”(open source way),比如透明治理方案、与利益相关者和开发群体的公开讨论,以及公平的规则制定程序。吸收这些开源元素,不仅有助于中国在国内的标准化工作,也有助于中国成为一个有责任心的“国际股东”,被视为一个值得信赖的参与者。没有这种信任,中国在海外扩大影响力的任何企图都会被讽刺为“新富阶层”的粗旷鲁莽。无论这种看法是否合理,是否公平,都无助于任何科技和技术的长期正确发展。

(请看此“中国的开源世界”系列的第一篇和第二篇。)

如果您喜欢所读的内容,请用email订阅加入“互联”。每周两次,新的文章将会直接送达您的邮箱。请在TwitterLinkedIn上给个follow,与我交流互动!