Jerry Chen & Jason Risch | VC Funding for the Cloud

Hi, everyone. Welcome to GrayMatter, the podcast from GrayLock, where we share stories from company builders and business leaders. I'm Heather Mack, head of editorial at GrayLock.

大家好，欢迎来到GrayMatter，这是由GrayLock出品的播客节目，我们分享企业创始人和商业领袖的故事。我是Heather Mack，GrayLock的编辑主管。

Today, we're bringing an update to our Castles in the Cloud project, which we launched last year to map the Cloud technology landscape and find opportunities for startups to compete with the big three-cloud providers of AWS, Google Cloud, and Microsoft Azure. We've compiled and analyzed data for 2021 in the beginning of 2022. These figures include the number of companies in the sector, the number of submarkets, and the amount of venture capital dollars invested in startups.

今天，我们将更新我们的“云堡垒”项目。去年我们推出了这个项目，旨在绘制云科技领域的地形图，为初创公司寻找与AWS、谷歌云和微软Azure三大云提供商竞争的机会。我们已于2022年初编制和分析了2021年的数据。这些数字包括该领域公司的数量、子市场的数量以及投资于初创公司的风险投资美元数量。

Today's episode is the audio version of our analysis of venture capital funding into cloud startups. GrayLock investors Jerry Chinn and Jason Rish wrote the essay, which you can find on the analysis section of the Castles in the Cloud project website at graylock.com/castles. Jason reads the essay.

今天的节目是我们对云创业公司风险投资资金分析的音频版本。GrayLock的投资者Jerry Chinn和Jason Rish写了这篇论文，您可以在graylock.com/castles项目网站的分析部分找到它。Jason朗读这篇论文。

You can find all GrayMatter content on the blog section of graylock.com and you can subscribe to GrayMatter podcasts on SoundCloud, Spotify, YouTube, or wherever you get your podcasts.

你可以在graylock.com的博客部分找到所有的GrayMatter内容，你也可以通过SoundCloud、Spotify、YouTube或其它提供播客的平台订阅GrayMatter的播客节目。

Diana Ramon the need for reliable tools that advance and protect the capabilities of cloud will continue to present opportunities for startups to thrive. The long-term trend of building cloud-native applications will continue to provide attractive startup opportunities for founders.

Diana Ramon认为，云计算可靠工具的需求将继续推动和保护云计算的能力，为初创企业提供繁荣的机会。长期建立云本地应用程序的趋势将为创始人提供有吸引力的初创企业机会。

We've compiled and analyzed new data on the sector as part of our Castles in the Cloud project, mapping the activity in the venture capital-funded startup ecosystem with that of the big three cloud providers, AWS, Azure and GCP. Total VC funding into the entire cloud sector hit 50.3 billion among tracked companies up from 15.3 billion in 2020.

我们在我们的“云中城堡”项目中编制和分析了有关该行业的新数据，将风险投资资助的初创企业生态系统中的活动与AWS、Azure和GCP等三大云服务提供商的活动进行了比较绘制。 追踪的企业整个云行业的风险投资总额达到了50.3亿美元，这一数字比2020年的15.3亿美元有所增加。

We discuss four markets drawing the most funding as well as one market edge we project to have more room to grow in the next section. These markets are security, AIML, management and governance, dev tools, edge. In context, the big three cloud providers have each expanded tremendously within the $212 billion market. AWS remains the market leader with a 33% share, followed by Microsoft Azure with 22% and Google with 10%. Individually, each company grew significantly with a 37% increase for AWS, 46% for Azure and 44% for GCP. Such growth rates are remarkable for a market of that size, demonstrating the cloud transition has not decelerated even as businesses move back to in-person operations.

在下一部分中，我们将讨论吸引最多融资的四个市场，以及我们预计在未来还有更多增长空间的一个市场。这些市场包括安全、人工智能/机器学习、管理和治理、开发工具和边缘计算。在这个背景下，三大云服务提供商都在价值2120亿美元的市场中实现了巨大的扩张。AWS依然是市场领导者，占据了33%的份额，其次是Microsoft Azure和Google分别占据了22%和10%的份额。分别来看，每家公司都实现了显著增长，AWS增长了37%，Azure增长了46%，GCP则增长了44%。这样的增长率在如此规模的市场中是值得注意的，说明即使企业开始回归线下运营，云服务的转型也没有减速。

Some analysts believe AWS could soon reasonably be considered a trillion dollar company if valued independently from the rest of Amazon. Taken together, we believe the tandem growth in private and public cloud companies demonstrate that there are still many opportunities for startups both competing with and collaborating with the major public companies in the sector.

一些分析师认为，如果从亚马逊其他业务中独立估值，AWS可能很快成为一家万亿美元的公司。总体来说，我们认为私有云和公共云公司的联合增长表明，在该领域竞争或合作的初创公司仍然有很多机会。

Top markets. Per our most recent data set, these four markets received the most VC funding. We believe there will be significant opportunities for startups and these sub-sectors going forward.

根据我们最近的数据集，这四个市场获得了最多的风险投资。我们相信未来初创企业和这些子行业将有显著的机会。

One, security. Increasing attacks draw increasing investment. While the cloud provides amazing benefits, it can also present increased security risk. Not only is the potential surface area for tax larger, it is also outside the safety of controlled on-prem systems and thus often poorly understood by traditional security teams and vendors. The number of cyber attacks per week on corporate networks increased 50% in 2021 compared to 2020, peaking at an all-time high in December due to the high profile log4J vulnerability.

一、安全性。攻击越来越多，投资也越来越多。虽然云计算带来了惊人的好处，但也可能存在更高的安全风险。潜在的攻击面积不仅更大，而且不受控制的本地系统的安全保障也更容易被攻击，因此常常被传统安全团队和供应商所忽视。2021年企业网络每周的网络攻击次数比2020年增加了50％，由于高调的log4J漏洞，在12月份达到历史最高水平。

Investment in security companies mirrored this frenzy of activity. Security now has the second highest number of tracked unicorns 18 in our database, with new unicorns raising across areas from cripple, log analysis, axonious asset management, panther, sim, and big ID, data classification. As we noted last year, security had one of the biggest jumps between 2019 and 2020 in aggregate funding. That trend continued this year, only more so.

对安全公司的投资反映了这股热潮式的活动。在我们的数据库中，安全现在是追踪到的独角兽数量第二高的领域，有18个新的独角兽在残疾人、日志分析、axonious资产管理、panther、sim、大数据ID数据分类等领域崛起。正如我们去年所述，安全在2019年和2020年的集体融资中跃升幅度最大之一。这种趋势在今年继续，而且只会更多。

Security startups received $7.5 billion from VC investors in 2021 compared with $2.1 billion in 2020. Within security, the leading subcategories were $3.03 billion for threats detection and $2.05 billion for vulnerability management, including monster rounds by Whiz, $500 million raised, Orca, $700 million raised, and Lacework, $1.83 billion raised. Identity and access management, a related market to security, also saw an additional $1.3 billion in funding.

2021年，安全初创企业从风险投资家那里获得了75亿美元的资金，而2020年只有21亿美元。在安全领域，主要的子类别是威胁检测（303亿美元）和漏洞管理（205亿美元），包括Whiz（筹集了5亿美元）、Orca（筹集了7亿美元）和Lacework（筹集了18.3亿美元）等公司的大额融资。身份和访问管理作为安全的相关市场，也获得了额外的13亿美元资金支持。

Security startup trends for 2022. Strong valuations will persist for security startups, even in rocky market conditions. The tailwinds behind security are such that both public and private companies remain strong in spite of today's economic volatility. Many public security companies have fared comparatively better than their software counterparts in the recent contraction of tech company multiples. For example, as of May 17, 2022, the NASDAQ tech 100 is down 29% year to date, while the NASDAQ CTA cybersecurity index is down about 50% less at down 20% year to date.

2022年安全创业趋势。即使在市场不稳定的情况下，安全创业公司的估值仍将保持强劲。安全背后的推动力很强，因此无论是上市公司还是私人公司，在今天的经济波动中都保持强劲。许多上市安全公司在技术公司倍增收缩的最近一波浪潮中表现比他们的软件同行要好。例如，截至2022年5月17日，纳斯达克科技100指数年初至今下跌了29％，而纳斯达克CTA网络安全指数跌幅则较少，年初至今下跌了20%。

Cybersecurity is often viewed as less discretionary than other categories of software spend, regardless of market conditions. For private companies operating in security, we expect financing to remain healthy. We saw multiple security companies raise strong rounds at higher valuations before the recent downturn. Abnormal security, which was incubated at Greylock, just raised $210 million and is now valued at $4 billion.

网络安全通常被视为比其他软件类别的支出更为必要，不受市场状况的影响。对于从事安全领域的私营公司，我们预计其融资将保持健康态势。在最近的经济下滑前，我们见证了多家安全公司以更高的估值成功进行了强劲融资。比如，起源于Greylock的Abnormal Security刚刚筹集了2.1亿美元，目前市值达到40亿美元。

Opportunity and specialization. Moreover, we expect opportunities for startups will abound. Generally speaking, the big cloud providers lack the security background and holistic view across a customer's entire environment to compete meaningfully in this area. While the big providers are taking steps to provide more options for customers, adding more security services than any other new services in 2021, these tools typically function in a supportive capacity for third party tools. This is where startups have the edge.

机遇与专业化。此外，我们预计初创企业的机会将充满可能性。总体而言，大型云服务提供商缺乏安全背景和全面的客户环境视图，无法在这个领域有意义地竞争。虽然大型供应商正在采取措施为客户提供更多选择，并在2021年增加了比任何其他新服务更多的安全服务，但这些工具通常是第三方工具的支持性能力。这就是初创企业的优势所在。

One of the biggest challenges we hear from CSOs is the sharp rise in the noise and identified security risks in vulnerabilities in a cloud environment. Combined with the shortage of qualified security professionals out there, security teams are overwhelmed and unable to correctly prioritize which ones to remediate first. Prioritization derived from business-specific context on which assets are most critical can help address this challenge, as can the rise of automation and security engineering. We believe there is significant opportunity for the small, hyper-focused teams at startups to solve this problem.

我们听到来自首席安全官的最大挑战之一是云环境中噪音和已知安全风险的急剧上升。加上业内缺乏合格的安全专家，安全团队不堪重负，无法正确地优先处理要解决的问题。通过业务特定背景下的优先级来确定哪些资产最为关键可以帮助应对这一挑战，而自动化和安全工程的提升也有助于解决问题。我们相信，初创公司小型高度专注的团队有巨大机会解决这个问题。

AIML. Customization and tying ML models to business value. As we wrote last year, the big three possess seemingly impenetrable modes, proprietary data sets combined with seamless interoperability and economies of scale. While these continue to be barriers, they have partially been offset by companies' need to customize AIML to their specific businesses and use cases. This has led AIML to become the category with the most unicorns in our database, 24, the second highest funding amount, 6.4 billion in 2021, and the most tracked companies, 51.

AIML. 定制化和将 ML 模型与业务价值相结合。正如我们去年所写的那样，大型企业拥有看似不可穿透的模式、专有数据集以及无缝互操作性和规模经济。尽管这些仍然是障碍，但它们已经部分被公司对 AIML 的特定业务和用例进行定制的需求所抵消。这使得 AIML 成为我们数据库中独角兽最多的类别，共有24家公司，2021年的募资金额也是第二高的，达64亿美元，并且追踪的公司最多，达到了51家。

Many of these unicorns are in core areas such as model development and training data. In this case, they are competing directly against the cloud providers end-to-end platforms like SageMaker, Azure ML, and Google AI platform. In model development, DataRobot has grown a large business by tying ML to business value and empowering technical analysts in addition to data scientists. Relative to the high number of tracked companies in the AIML category, other parts of ML ops have developed more slowly. This is because just a handful of companies opt to purchase a best-in-breed ML stack rather than building their own or purchasing an all-in-one platform. However, we anticipate more companies will be swayed by the sophistication and usability of emerging tools such as those offered by base 10.

这些“独角兽”中许多是在核心领域，如模型开发和训练数据。在这种情况下，它们直接与云供应商的端到端平台如SageMaker、Azure ML和Google AI平台竞争。在模型开发方面，DataRobot通过将ML与业务价值联系起来，并赋予技术分析专家以及数据科学家更多的权力而成长为一个庞大的业务。相对于AIML类别中跟踪的高数量的公司，ML操作的其他部分发展得更慢。这是因为只有少数公司选择购买最佳的ML堆栈，而不是自己构建或购买一体化平台。但是，我们预计更多的公司将被新兴工具的复杂性和易用性所吸引，例如由base 10提供的工具。

The applied ML company, which recently launched with 20 million in series A funding led by Greylock, makes models easy to deploy by surrounding them with business logic and front-end UI, so businesses can quickly incorporate models into business critical processes, and the workflows of domain experts. Other companies like HuggingFace, which recently raised $100 million, and Wates and Viasis have developed engaged communities, which will ultimately compete against the big three clouds on quality of ML practitioners. Unicorns also continue to grow in domain-specific areas, while the cloud providers can provide off-the-shelf APIs, many companies struggle with their flexibility and customizability to adapt them to their own data and use cases.

这家应用机器学习技术的公司前不久刚刚获得了2000万美元的A轮融资，这笔资金由灰鸟领投。公司通过将模型与业务逻辑和前端UI相结合，使其易于部署，从而使企业能够快速将模型纳入业务关键流程和领域专家的工作流中。像HuggingFace这样最近筹集了1亿美元的公司以及Wates和Viasis，已经发展了积极的社区，并最终将根据机器学习从业者的素质与三大云计算公司竞争。独角兽也在领域特定领域不断成长，而云服务提供商虽然可以提供现成的API，但许多公司在灵活性和可自定义性方面仍然面临困境，以适应自己的数据和用例。

Companies like Riscified and Fortur have built fraud detection algorithms and workflow and e-commerce, while InstaVase is a unicorn in document extraction with vertical solutions like M&E. Solutions like Medical Claims Processing and Lone Processing.

像Riscified和Fortur这样的公司已经构建了针对电子商务的欺诈检测算法和工作流程，而InstaVase则是文件提取中的独角兽，在M&E等垂直解决方案中提供医疗索赔处理和单独处理等解决方案。这些公司的解决方案可以帮助企业识别欺诈行为，并提高效率和准确性。

ML Trends in 2022. The rise of large language models. Startups developing NLP tools are also among the highest number of unicorns, five of any segment.

2022年的机器学习趋势：大型语言模型的崛起。开发自然语言处理工具的初创公司也是最多的独角兽之一，高达五家。

In the last decade, most of the acclaim and funding want to startups made possible by advances in vision technology, and autonomous vehicles and extra interpretation have dominated headlines.

在过去的十年里，由于视觉技术的进步和自动驾驶车辆以及额外的解释的发展，大多数赞誉和资金都流向了初创企业，这些成就在头条新闻中占据主导地位。

Recently, however, the rise of large language models like Bird from Google and GPT-3 from OpenAI have driven increased funding and interest in NLP, and companies that can be generated from it. GitHub Copilot is a prominent example. According to GitHub's internal benchmarks, 30% of new code is being suggested by Copilot in some programming languages.

近年来，如Google的Bird和OpenAI的GPT-3这样的大型语言模型的崛起，推动了对自然语言处理(NLP)的越来越多的资金和兴趣投入，以及可以从其中产生的公司。GitHub Copilot就是一个突出的例子。据GitHub内部基准测试，Copilot在一些编程语言中建议了30%的新代码。

We're excited to see new vertical and horizontal businesses built on top of these large language models, which may be either developed by the company itself or built on top of providers like OpenAI. We're encouraged by the swift progress in the field we've seen in promising NLP startups like Adapt, which just landed 65 million in Series A funding and Anthropic, which recently announced its 580 million Series B.

我们很兴奋地看到这些大量语言模型上面构建的新型垂直和水平业务，这些业务可能是由公司自己开发，也可能是建立在像OpenAI这样的提供者之上。我们对在这一领域中看到的有希望的自然语言处理初创公司的迅速进展感到鼓舞，例如刚获得6500万美元A轮融资的Adapt和最近宣布5800万美元B轮融资的Anthropic。

Acceleration of startup activity. It is likely more difficult to compete with the cloud providers and ML hardware and compute, but some startups are competing in related areas given the explosion of costs associated with training and deploying large models. OctoML, building off of TBM and modular AI, building off of MLIR, are developing tools that leverage compilers and runtime to optimize deployment of hardware. Mosaic ML is applying best practices across training workflows, training algorithms, and hardware runtimes to dramatically cut training costs.

初创企业活动的加速。由于云服务提供商和机器学习硬件和计算的竞争可能更加困难，但考虑到培训和部署大型模型所需的成本激增，一些初创企业正在相关领域进行竞争。OctoML基于TBM和模块化AI构建，建立在MLIR之上，并开发了利用编译器和运行时优化硬件部署的工具。Mosaic ML将最佳实践应用于培训工作流程、培训算法和硬件运行时，以大幅降低培训成本。

We see continued opportunity for startups to come in as third parties that can innovate down the ML infrastructure stack and move beyond the existing focus at the model data and application level.

我们看到，初创企业有持续的机会作为第三方进入机器学习基础设施层，进行创新，并超越目前关注的模型数据和应用层面。

Management and governance. Giving businesses control and accountability. Tools that enable companies to extend control over their cloud drew considerable investor interest last year. Funding into the category reached 3.4 billion in 2021, a threefold increase from 2020. Such tools are in line with a few trends we foresee increasing in 2022.

管理和治理。为企业提供控制权和责任。能够使公司扩展对其云的控制的工具去年引起了投资者的极大兴趣。该类别的投资达到了 34 亿美元，比 2020 年增长了三倍。这些工具符合我们预计在 2022 年会增加的一些趋势。

More sophisticated data management and protection. As the risk and actual occurrence of cyber attacks has risen, cyber insurance providers such as startup unicorns coalition and a bay have become mandatory tools for businesses. However, the market is hardening as the sheer volume of attacks has led to an increase in premiums due to unexpectedly high losses.

更复杂的数据管理和保护。由于网络攻击的风险和实际发生率上升，像联盟和盾的创业独角兽这样的网络安全保险提供商已成为企业必不可少的工具。然而，随着攻击数量的增加导致预期外的高损失，市场正在变得越来越困难，保险费用也在逐渐上涨。

Ransomware attacks in particular are creating a spiral in which companies use their insurance policies to pay the hackers to recover their data, which increases the incentive for hackers in the first place. Companies like Rubric, Cohesity and VM help to mitigate this risk in the first place by providing backup and disaster recovery, making assets recoverable without paying and depending on the hackers.

勒索软件攻击特别是创造了一个恶性循环，公司为了恢复他们的数据，使用他们的保险政策来支付黑客，这增加了黑客攻击的动机。像 Rubric、Cohesity 和 VM 这样的公司能够在第一时间减轻此风险，他们提供备份和灾难恢复，使资产能够在不支付和依赖黑客情况下恢复。

Rising ESG expectations. ESG has become a board level priority in the enterprise over the past few years, given predictions around increased government regulation, pressure from large shareholders like BlackRock and consumer expectations. Enterprises are working to measure, analyze and offset carbon emissions data to meet net neutrality objectives and the cloud plays a significant role.

越来越高的ESG期望。过去几年，由于政府监管增加的预测、BlackRock等大型股东的压力以及消费者期望，ESG已成为企业董事会的优先事项。企业正在努力衡量、分析和抵消碳排放数据，以实现净中立目标，而云计算发挥着重要作用。

AWS Carbon Footprint Tool, GCP Carbon Footprint and Microsoft Emissions Impact Dashboard all help customers explore the environmental impact of their cloud usage. Salesforce and Microsoft have each also launched sustainability cloud products aimed at helping customers analyze emissions data across their entire organization and competing with independent companies like Watershed and Persephone. We believe this will continue to be an area of competition in 2022 both to serve these markets and to make cloud services as efficient as possible.

AWS碳足迹工具、GCP碳足迹和微软排放影响仪表板都有助于客户探索其云使用对环境的影响。Salesforce和微软也分别推出了可持续云产品，旨在帮助客户分析整个组织的排放数据，并与像Watershed和Persephone等独立公司竞争。我们相信，这将在2022年继续成为竞争领域，既服务这些市场，又尽可能地使云服务更加高效。

Cost Management. Cloud Cost Management was the number one ranked cloud initiative overall among responders to flex areas 2022 state of the cloud report. Respondents reported being an average of 12% over budget and self-estimating 32% of overall cloud spend was wasted. FinOps teams are increasingly working with engineers to create better predictability as cost becomes one of the three key components of cloud management along with performance and security.

成本管理。云成本管理是 flex areas 2022 云状态报告中应答者最高排名的云计划。应答者报告称，他们的预算超支平均为12％，总体云花费的32％被浪费。FinOps 团队越来越多地与工程师合作，以创建更好的可预测性，因为成本已成为云管理的三个关键组成部分之一，连同性能和安全性。

Our database lists eight predominantly early stage companies providing products to help customers estimate and control cloud costs by properly sizing deployments. We believe bridging the gap between finance and development is key to enabling engineers to make accurate and informed decisions in real time.

我们的数据库列出了八家主要处于早期阶段的公司，提供产品来帮助客户通过适当大小规划来估算和控制云成本。我们认为，弥合财务和开发之间的差距是让工程师能够实时做出准确和明智决策的关键。

As companies see ballooning bills we expect to see this concept spread into other areas. Cribble provides the service and the observability stack by filtering and transforming logs in front of downstream products like Splunk. Several stealth companies we have seen are looking to apply a similar concept in analytical databases. Dev tools.

随着公司的费用不断增长，我们预计这个概念将在其他领域得到推广。Cribble通过在下游产品（如Splunk）前过滤和转换日志，提供服务和可观察性栈。我们发现，一些秘密的公司正在寻求在分析数据库和Dev工具等领域应用类似的概念。

Extending engineering resources. 2021 was a banner year for the emergence of large DevOps players. As CircleCI, CloudBees, Harness and LaunchDarkly all become DevOps Unicorns while HashieCorp and GitLab went to public. 2021 saw customers adopt third party DevOps platforms and mass. CICD is traditionally an area where engineers have been very opinionated and prefer to rely on open source software in order to have complete control over their stack, both from a customization standpoint as well as a reliability standpoint. However, the cloud has driven companies to become comfortable with the idea of strafting away mission critical infrastructure, even in areas like their build pipeline. Given expensive engineering resources and the talent and crunch, many companies are also aiming to focus as much of their engineering talent as possible on building products instead of internal build processes. The convergence of these trends has created an opportunity for these DevOps companies to scale in an area that has been historically hard to monetize.

扩展工程资源。2021年是大型DevOps参与者崛起的标志性一年。随着CircleCI、CloudBees、Harness和LaunchDarkly成为DevOps独角兽，HashieCorp和GitLab走向公众。2021年，客户大量采用第三方DevOps平台。CICD传统上是工程师非常有意见的领域，他们更倾向于依赖开源软件，以便从定制和可靠性两个方面完全控制他们的堆栈。然而，云驱动企业变得对在诸如构建流水线等关键基础设施方面采取改进为常态。鉴于昂贵的工程资源、人才和时间紧缺，许多公司也力图尽可能将他们的工程人才集中在构建产品上，而不是内部构建流程上。这些趋势的融合为这些DevOps公司提供了在历史上难以实现盈利的领域中扩大规模的机会。

Dev tool trend for 2022. Bundling and unbundling. CircleCI, Harness and LaunchDarkly all started as different point solutions. CircleCI and continuous integration, Harness and continuous deployment, and LaunchDarkly and feature flagging. However, these and other DevOps companies have aimed to move up or down the stack to create an end to end solution. Many customers are increasingly open to adopting platforms in CICD to ensure strong interoperability and simplicity. On the flip side, we also see the opportunity to unbundle some other parts of Dev tools like GitHub as will be the subject of a future essay.

2022年的开发工具趋势：打包和解包。CircleCI、Harness和LaunchDarkly最初都是不同的解决方案。CircleCI提供持续集成、Harness提供持续部署，而LaunchDarkly则提供功能标志。然而，这些以及其他DevOps公司都致力于向上或向下移动堆栈，以创建一个端到端的解决方案。许多客户越来越愿意采用CICD平台，以确保强大的互操作性和简单性。另一方面，我们也看到了解包Dev工具的其他部分的机会，例如将是未来一篇文章的主题的GitHub。

Edge. The next cloud frontier. One smaller market in which we expect to see significant growth and innovation is an edge computing. Edge computing consists of storage and compute on the edge and is a powerful enabler for some of the other cloud markets we track, such as gaming and IoT. Gaming as a whole recently became the largest media category, period, and the demand for instant response times is huge in cloud gaming. The rise of AI models deployed on the edge and devices like cameras is also driving demand for edge computing. Another form of edge computing takes place in the browser, where technology like WebAssembly, Wasm, are bringing impressive capabilities. Wasm allows compiled languages like C++ and Rust to be packaged into a binary and loaded into a browser. Companies like Figma have demonstrated the power of Wasm over the past few years by significantly cutting low times and latency in their browser-based applications. Leading companies like Shopify are also implementing Wasm outside of the browser, and CloudFlare is investing heavily in WebAssembly on CloudFlare workers. Despite the power of this technology, we have not seen an independent unicorn in the edge space yet. We predict that this will change in the next few years as companies are started that make Wasm easier to use and others continue to harness Wasm for applications.

边缘计算是下一个云计算领域的重要市场之一，我们预计在这个市场中会出现显著的增长和创新。边缘计算包括在边缘上的存储和计算，是其他云计算市场的强大推动力，如游戏和物联网。整个游戏行业最近成为了最大的媒体类别，对于云游戏而言，即时响应时间的需求非常大。边缘上部署的人工智能模型和相机等设备的增加也推动了对边缘计算的需求。另一种形式的边缘计算是在浏览器中进行的，其中像WebAssembly（Wasm）这样的技术带来了令人印象深刻的功能。Wasm允许编译语言如C++和Rust打包成二进制并加载到浏览器中。像Figma这样的公司已经通过Wasm在过去几年中展示了Wasm的强大功能，极大地减少了其基于浏览器的应用程序的低延迟。领先的公司如Shopify也在浏览器外实现Wasm的应用，CloudFlare正在大力投资于CloudFlare workers上的WebAssembly。尽管这项技术很强大，但我们还没有看到具有独立地位的边缘独角兽公司。我们预测，在未来几年中，随着开始创立能够更易于使用Wasm的公司以及其他公司继续利用Wasm进行应用开发，这种情况将发生改变。

2022 preview is a very important step to the development of the cloud. We have seen a very important step to the development of the cloud. We have seen a very important step to the development of the cloud. As data from Cardo shows a 25% decrease in average series A valuations and a 42% decrease in series C valuations. However, seed funding valuations remain strong, and the best companies of all stages will continue to grow as opportunity of bounds to compete against the Cloud Big 3 given the secular growth of the overall cloud market. COVID-19 and work from home accelerated the cloud transition, and while many consumer and commerce stocks buoyed by COVID-19 have seen recent declines, there are no signs of the enterprise cloud transition reverting or decelerating. Many stand out private cloud companies have the scale and predictability to go public and will do so once the public market settle. We remain optimistic and are continuing to invest aggressively in the best founders building in the cloud opportunities above.

2022的前瞻对云的发展是非常重要的一步。从 Cardo 的数据中可以看出，A轮平均估值下降了25%，C轮下降了42%。然而，种子轮融资估值仍然强劲，而各阶段的最佳公司将继续成长，随着总体云市场的长期增长，有着大量机会与云三巨头竞争。COVID-19 和远程工作加速了云转型，虽然许多受COVID-19 支持的消费和商业股票最近出现了下跌，但没有迹象表明企业云转型会逆转或减速。许多出色的私有云公司具有规模和可预测性，可以上市，并在公共市场稳定后实现上市。我们仍然持乐观态度，并继续积极投资于在云机会上打造的最佳创始人。

That concludes this episode of Gray Matter. You can find this essay and all castles in the cloud related content at graylock.com slash castles. You can find all other gray matter content under the blog section of graylock.com, and you can subscribe to the podcast on SoundCloud, Spotify, YouTube, or wherever you get your podcasts. I'm Heather Mack. Thanks for listening.

这一集的《灰色物质》到此结束了。如果您想阅读这篇文章或是找到与云彩中的城堡相关的内容，请前往graylock.com/castles。如果您想查阅其他的《灰色物质》内容，请前往graylock.com的博客区块。您可以通过SoundCloud、Spotify、YouTube或任何其他您可以获取播客的地方订阅该播客。我是Heather Mack。感谢您的收听。