Lovable signs multiyear deal with Google Cloud to up usage 5x, source says Lovable 与 Google Cloud 签署多年期协议，将使用量提升5倍，消息人士称

Google just quietly revealed its hand in the high-stakes game of artificial intelligence portfolio management, and the move is as cynical as it is brilliant. By inking a massive expanded cloud deal with Stockholm’s Lovable, a “vibe-coding” startup, Google isn’t just securing a customer; it’s actively engineering the success metrics for its billion-dollar bet on Anthropic.

The facts are straightforward. Lovable, which lets non-experts build software through natural language, is a rocketship. It claims to have hit a $400 million annualized run rate in February, adding $100 million in a single month with a team of just 146 people. Over half of the Fortune 500 uses it. So, naturally, it needs a bigger cloud engine. Under the new multiyear agreement, it will quintuple its consumption of Google Cloud, which includes heavy usage of AI models. This is where it gets interesting. Lovable gets expanded access not just to Google’s homegrown Gemini, but also to Anthropic’s Claude, the current darling for coding tasks.

This is not a neutral procurement decision. It is a strategic financial maneuver. Remember, in April, Google invested $10 billion in Anthropic, with a promise of another $30 billion contingent on performance targets. That deal was struck at a $350 billion valuation. A month later, Anthropic raised at a valuation nearing $1 trillion. The terms of Google’s investment are now public leverage. To hit those performance targets, Anthropic needs massive, verifiable commercial usage of its models. And here comes Lovable—one of Europe’s fastest-growing tech companies ever—being handed a direct pipeline to become a colossal consumer of Claude, courtesy of a Google-funded discount.

The circle is complete. Google uses its balance sheet to invest in Anthropic, then uses its cloud sales arm to guarantee Anthropic a marquee customer, thereby helping Anthropic hit the milestones that trigger Google’s further funding. It’s a self-reinforcing loop of capital and consumption, a masterpiece of financial engineering disguised as a partnership. Lovable benefits with best-in-class tools and infrastructure. Anthropic gets a usage graph that rockets upward. And Google? It gets to count both the cloud revenue and the strategic success of its Anthropic investment. It’s hedging and synergy in one clean stroke.

This also lays bare the precarious reality of the "model-agnostic" cloud promise. Google Cloud is selling access to a direct competitor’s product—Anthropic’s Claude—alongside its own Gemini. This isn’t a philosophical commitment to openness; it’s a cold-blooded market calculation. If the market overwhelmingly prefers Claude for coding (which it currently does), Google Cloud can’t afford to not offer it. Better to monetize the trend and keep the customer on GCP than to lose them entirely. The cloud war is no longer just about who has the best infrastructure; it’s about who can broker the best access to the fragmented universe of AI models.

Lovable itself is a fascinating case study in the new, brutally efficient startup model. To achieve that revenue with that headcount is a testament to the power of product-led growth amplified by AI. But it also raises a eyebrow. "Vibe-coding" sounds like a toy, yet it’s generating half a billion dollars in annualized revenue. This suggests the market for lowering the barrier to software creation is even more vast and desperate than anyone imagined. Every marketing executive, small business owner, and product manager now believes they can build their own tools. Lovable is selling pickaxes in a gold rush, and Google is ensuring it has an unlimited supply of the sharpest ones.

Critically, one must wonder about the sustainability and the substance beneath these sky-high numbers. Annualized run rate based on a single month’s spike is a vanity metric prone to collapse. Is this usage driven by genuine, sticky enterprise adoption, or by a wave of experimentation fueled by easy money and hype? The "over half of Fortune 500" claim is equally nebulous. Are these companies paying for full-blown, mission-critical deployments, or are they tinkering with free trials and pilot projects? The partnership announcement is designed to project irrefutable momentum, but the real test is whether these customers are building on Lovable’s platform as a core competency or just as a curiosity.

Ultimately, this deal is less about two companies collaborating and more about the centralization of AI power. The ecosystem is rapidly stratifying into a few hyperscalers (Google, Microsoft) and a few foundational model makers (OpenAI, Anthropic, Google DeepMind). Startups like Lovable become the vital conduit, the application layer where these raw AI capabilities are converted into revenue. But their destiny is tied to the deals their providers make in the boardroom. Today, Lovable is the partner of choice. Tomorrow, if the financial calculus shifts, its access to the best models could be throttled or repriced.

Google is playing chess while others are playing checkers. It is using its cloud division not just to sell compute, but to strategically populate and validate its AI investment portfolio. Lovable is the perfect vehicle: fast-growing, model-hungry, and just successful enough to make Anthropic’s usage stats look phenomenal. This is the new playbook. In the age of AI, the most powerful tech companies aren’t just building products; they’re building and managing entire ecosystems of dependency, where every partnership, every cloud deal, and every investment is a move in a much larger game of capital concentration. The vibe may be casual, but the strategy is utterly, ruthlessly corporate.

斯德哥尔摩的“vibe-coding”新贵Lovable与谷歌云宣布扩大合作，这笔交易表面上是一次顺理成章的云服务扩容——Lovable在其平台上的足迹将扩大五倍，并能更广泛地使用Anthropic的Claude与谷歌自家的Gemini模型。但拨开这层技术合作的外衣，我们看到的更像一出精心编排的“北欧联姻”：谷歌正试图用资本和算力，将一家欧洲的增长怪兽、其重金押注的AI模型公司，以及自身云平台捆成一个命运共同体。

交易的细节很耐人寻味。它没有披露金额，但据称涉及五倍的云使用量增长，这本身就是个骇人的数字。要知道，Lovable在2024年2月宣称年化收入已突破4亿美元，单月增长就达到1亿美元，而员工只有146人。如果这些数据属实，其人均产出已达到令人瞠目的程度，它自称超过一半的财富500强企业在使用其产品。这种增长速度和效率，让它看起来不像是一家传统的软件公司，而更像一个由AI驱动的“增长引擎”，每一行代码都在直接兑换成惊人的ARR（年度经常性收入）。谷歌此次加深合作，无异于给这台引擎加注更高级别的燃料——同时来自Claude和Gemini的AI能力。

但这背后真正的棋局，显然与谷歌对Anthropic那笔庞大而复杂的投资紧密相连。谷歌在四月份以3500亿美元的估值向Anthropic投资了100亿美元现金加算力，并承诺若达成特定目标，将再追加300亿美元。而就在一个月后，Anthropic自身的融资估值已飙升至近1万亿美元。这笔与Lovable的合作，正是帮助Anthropic达成那些“性能目标”的关键一步。Lovable作为欧洲增长最快的科技公司之一，其海量的编码任务使用量，将直接转化为对Claude模型的调用次数和数据反馈。换句话说，谷歌正在用Lovable的业务增长，来“喂养”自己巨额投资的标的公司，确保那笔可能高达400亿美元的后续投资能够顺利触发。这是一套完美的资本逻辑闭环：谷歌投钱给模型公司，模型公司赋能应用公司，应用公司的增长又反过来证明模型公司的价值，并帮助达成对赌条款，最终巩固谷歌在整个AI生态中的霸主地位。

然而，当我们跳出这精巧的架构设计，审视行业现状时，一股浓重的“AI泡沫”气息扑面而来。Lovable所代表的“vibe-coding”赛道，本质上是将AI从辅助工具升级为生产主角，它承诺用自然语言生成应用，极大地降低了开发门槛。这种模式的爆发性增长，恰恰印证了市场对“AI生产力”的狂热追逐。但问题在于，这种由资本疯狂灌溉出来的、近乎指数级的增长，其可持续性要打上一个大大的问号。146人的团队支撑4亿美元的年化收入？这数字听起来更像一个精心设计的增长故事，而非可持续的商业模式。当风口略微转向，或者客户为AI付费的狂热消退时，如此轻资产、高杠杆的结构是否还能维持？它更像一个建立在AI算力悬崖边的精致沙堡。

更辛辣地看，这场合作也折射出当前AI竞赛的一种集体浮躁。巨头们不再满足于单纯的技术竞逐，而是迫不及待地通过错综复杂的投资与合作协议，编织一张将客户、模型、平台绑定在一起的巨网。Lovable、Anthropic、谷歌，三者形成了“应用-模型-平台”的铁三角。这种深度捆绑固然能产生协同效应，但也极大地增加了生态的脆弱性和同质化风险。所有人都在追逐下一个“十亿用户”的AI杀手应用，但真正扎实的、解决根本问题的创新，是否被这些令人眼花缭乱的资本合纵连横掩盖了？我们看到的不是百花齐放，而是资本逻辑驱动下的生态联盟竞赛，比拼的是谁能更快地画出那条陡峭的增长曲线。

所以，Lovable与谷歌的这次扩表，远不止是一笔云采购合同。它是AI资本游戏的一个标准模版：增长神话需要顶级算力支撑，顶级算力需要模型实力背书，模型实力又需要增长神话来验证价值。三者彼此依存，互相抬轿。对于身处其中的146名Lovable员工而言，他们或许正享受着“全员高管、全员销售”的极致效率狂欢；但对于整个行业而言，我们或许正在见证一种由资本催熟的、高风险高回报的“AI资本主义”新模式。当潮水退去，除了那五倍扩张的云足迹和一串令人眩晕的财务数字，我们还能剩下什么真正经得起推敲的技术沉淀？这个问题，恐怕比这份合作协议上的任何一个条款都更加重要。