广告
加载中

3个90后创业4年 公司估值2000亿

黎曼 2026-08-31 09:51
黎曼 2026/08/31 09:51

邦小白快读

EN
全文速览

本文介绍了3名90后创办的AI初创公司Perplexity AI的发展历程,当前该公司完成新一轮融资后估值将超300亿美元(约合人民币2000亿元),成立不到四年估值翻了200倍,是AI搜索赛道成长最快的明星公司。

1. 核心业务与模式:公司主打AI对话搜索,能直接给用户输出提炼好带出处的答案,后续推出可自动执行电脑复杂任务的AI智能体产品;走差异化轻资产路线,不自研基础大模型,专注做检索和模型编排,调用多家厂商模型完成任务,迭代速度快。

2. 增长数据亮眼:推出AI智能体后不到八个月,年化营收从不足2.5亿美元攀升至超7.5亿美元,翻了近三倍;月活跃用户最高超1亿,包含数万家企业客户,A轮投资人获得了200倍回报。

3. 可参考的创业经验:抓住AI浪潮下用户未被满足的需求,差异化切入巨头垄断的赛道,小团队也能实现快速跨越式增长。

本文呈现了AI赛道初创品牌Perplexity AI的崛起路径,能给AI领域品牌经营者提供多方面参考。

1. 消费趋势洞察:用户对搜索服务的需求已经从传统的返回链接转向直接获取提炼完成的精准答案,对AI智能体自动执行复杂任务的需求增长迅猛,新的需求催生了新品牌的成长空间。

2. 差异化品牌定位经验:避开和谷歌、OpenAI等巨头拼自研大模型的重资产竞争,聚焦自身擅长的检索与模型编排环节,轻资产快迭代,快速切中用户痛点,成功将谷歌逼出舒适圈,打开了市场知名度。

3. 品牌营销与生态玩法:邀请顶级足球明星C罗投资并担任全球品牌代言人,借助名人流量快速打开全球认知;同时绑定头部产业资本英伟达,加入其Nemotron开源模型生态,既解决了第三方模型依赖的潜在风险,也进一步提升了品牌估值与行业影响力。

Perplexity AI的快速崛起,给AI领域创业者、从业者带来了明确的机会提示与风险预警,有很多可学习的经验。

1. 市场机会:AI搜索、AI智能体是当前增速最快的赛道,传统搜索引擎无法满足用户直接获取精准答案的需求,巨头尚未完全覆盖市场,初创团队仍有切入机会。

2. 可复制的商业模式:对于资源有限的初创团队,不自研基础大模型,采用“多模型调用+检索增强生成”的轻资产模式,聚焦用户体验迭代,能快速推出产品验证需求,实现增长。

3. 风险提示:依赖第三方大模型的模式存在供应链风险,一旦第三方模型涨价或限制访问,企业成本结构会直接承压,需要提前绑定稳定的生态合作伙伴规避风险。

4. 资本机会:当前英伟达等头部芯片厂商采用“股权投资绑定芯片采购”的模式布局AI应用层,优质的高增长AI应用项目更容易获得资本加持。

AI产业的快速发展,给科技制造、配套服务类工厂带来了新的商业机会与转型启示。

1. 产品需求变化:AI搜索尤其是AI智能体的爆发,带来了算力需求的高速增长,Perplexity转型AI智能体后已经成为算力大户,对芯片、算力基础设施的需求持续提升,给芯片生产、配套硬件制造工厂带来了大量稳定的订单需求。

2. 新商业机会:英伟达当前正在大力搭建Nemotron开源AI生态,吸纳了Perplexity等一众头部AI应用企业作为核心节点,需要大量硬件生产、技术配套的支持,给相关工厂带来了新的合作机会。

3. 数字化转型启示:AI浪潮下,传统工厂可以加快推进数字化转型,对接头部AI企业的生态需求,切入AI产业链配套环节,绑定核心生态玩家,获得新的增长曲线,摆脱传统业务的增长瓶颈。

对于AI领域相关服务商来说,本文透露了大量行业趋势、客户痛点相关的干货内容。

1. 行业发展趋势:当前AI产业已经从基础大模型的竞争,逐渐转向应用层落地的竞争,AI搜索、AI智能体是当前增长最快的应用方向,获得了产业资本的高度认可,未来增长空间巨大。

2. 客户痛点洞察:大量初创AI公司没有足够的资金和能力自研基础大模型,迫切需要轻资产切入AI赛道的解决方案,用户对搜索体验升级也有明确的未被满足的需求。

3. 业务方向参考:服务商可以围绕多模型路由调度、检索增强生成开发配套工具与解决方案,服务大量中小AI初创客户;同时,AI智能体的爆发带动了算力需求的持续增长,算力调度、配套技术服务等领域也会迎来持续的需求增长,服务商可以加入头部厂商的开放生态获取稳定客户。

Perplexity AI的成长和英伟达的生态布局,给AI领域平台商带来了生态建设、运营管理方面的诸多启示。

1. 洞察AI应用企业对平台的核心需求:多数AI应用层企业都需要稳定的算力供应和模型资源支持,依赖第三方模型的应用企业存在明显的供应链风险,平台企业可以通过生态绑定解决客户痛点,同时构建自身竞争壁垒。

2. 可参考的平台布局模式:英伟达“股权投资+生态绑定+芯片销售”的模式很值得借鉴,通过投资高增长AI应用企业,带动自身芯片和算力产品的销售,同时搭建开放生态吸引更多玩家加入,形成正向循环。

3. 招商与风险规避:AI搜索、AI智能体赛道的企业增长快、算力需求大,是平台优质的招商目标,可以重点对接;但布局过程中需要警惕估值泡沫,要关注被投企业的真实用户需求和营收增长,避免资本推高的虚假需求带来的风险。

本文记录了AI搜索赛道最新的产业发展动态,为产业研究提供了鲜活的案例和明确的研究方向。

1. 产业新动向:AI应用层已经跑出了独立的高增长头部玩家,成立不到四年的Perplexity AI估值突破300亿美元,成为AI搜索赛道估值最高的独立玩家,AI产业竞争已经从基础大模型转向应用层落地,AI智能体成为应用层新的增长催化剂。

2. 新商业模式研究:“轻资产多模型路由+检索增强生成”的商业模式得到了市场和资本的验证,打破了只有自研基础大模型才能在AI赛道成功的行业认知,为研究AI初创企业成长路径提供了新样本。

3. 值得研究的新问题:英伟达“股权投资绑定芯片采购”的布局模式存在隐忧,是否会催生行业估值泡沫,依赖第三方大模型的应用层企业如何解决供应链风险,这些都是AI产业发展中出现的新问题,值得深入研究。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

This article outlines the growth of Perplexity AI, a generative AI startup founded by three Gen Z entrepreneurs. After its latest funding round, the company is set to hit a valuation of over $30 billion (roughly 200 billion RMB). In less than four years since its founding, its valuation has increased 200-fold, making it the fastest-growing star company in the AI search track.

1. Core business and model: Perplexity’s flagship product is conversational AI search, which directly delivers users summarized answers with cited sources. It later launched an AI agent product capable of automatically executing complex tasks on computers. The company pursues a differentiated asset-light strategy: instead of developing its own foundational large language model (LLM) in-house, it focuses on information retrieval and model orchestration, calling on models from multiple providers to complete tasks, enabling much faster iteration.

2. Impressive growth metrics: Less than eight months after launching its AI agent, Perplexity’s annualized revenue surged from less than $250 million to over $750 million, a nearly threefold jump. Its monthly active users peaked at over 100 million, including tens of thousands of enterprise clients, and its Series A investors have already earned 200x returns.

3. Key entrepreneurial takeaways: By capitalizing on unmet user demand amid the AI boom and entering a giant-dominated sector through a differentiated strategy, small teams can still achieve rapid, leaping growth.

This article breaks down the rise of AI startup Perplexity AI, offering multiple insights for brand builders in the AI space.

1. Consumer trend insight: User demand for search has shifted from traditional link results to direct, summarized, accurate answers. Demand for AI agents that can automatically complete complex tasks is growing rapidly, and these unmet needs have created significant room for new brands to grow.

2. Differentiated brand positioning: Perplexity avoided the capital-intensive race of building in-house foundational LLMs with giants like Google and OpenAI. Instead, it focused on its core strengths in retrieval and model orchestration, running an asset-light operation with fast iteration to quickly address user pain points. The strategy has forced Google out of its comfort zone and earned Perplexity significant market recognition.

3. Brand marketing and ecosystem strategy: Perplexity secured investment from global football superstar Cristiano Ronaldo, who also serves as its global brand ambassador, leveraging celebrity influence to rapidly build global awareness. It has also partnered with leading industry player NVIDIA, joining NVIDIA’s open-source Nemotron model ecosystem. This move mitigates the potential risk of relying on third-party models, while further boosting Perplexity’s brand valuation and industry influence.

The rapid rise of Perplexity AI delivers clear opportunities and risk warnings for AI entrepreneurs and practitioners, with many actionable lessons.

1. Market opportunity: AI search and AI agents are currently the fastest-growing segments. Traditional search engines cannot meet user demand for direct, accurate answers, and incumbents have not yet fully covered the market, leaving room for early-stage teams to enter.

2. Replicable business model: For resource-constrained early-stage teams, the asset-light model of "multi-model orchestration + retrieval-augmented generation" (RAG), with no in-house foundational LLM required, allows teams to focus on iterating user experience, launch products quickly to validate demand, and achieve rapid growth.

3. Risk warning: The model of relying on third-party LLMs carries supply chain risk. If third-party models raise prices or restrict access, the company’s cost structure will face immediate pressure. Teams need to lock in stable ecosystem partners in advance to mitigate this risk.

4. Capital opportunity: Leading chipmakers like NVIDIA currently use a "equity investment tied to chip procurement" model to布局 the AI application layer. High-quality, fast-growing AI application projects are much more likely to secure capital backing.

The rapid growth of the AI industry has brought new business opportunities and transformation insights for technology manufacturing and supporting service factories.

1. Shifting product demand: The boom of AI search, especially AI agents, has driven explosive growth in demand for computing power. After shifting focus to AI agents, Perplexity has become a major computing power consumer, with continuously rising demand for chips and computing infrastructure. This has generated large, stable order demand for chip manufacturers and supporting hardware factories.

2. New business opportunities: NVIDIA is currently aggressively building out its open-source Nemotron AI ecosystem, with Perplexity and other leading AI application companies as core nodes. The ecosystem needs extensive support for hardware production and technical supporting services, creating new partnership opportunities for relevant factories.

3. Digital transformation insight: Amid the AI wave, traditional factories can accelerate digital transformation, align with the ecosystem needs of leading AI companies, enter AI industry supply chain segments, and partner with core ecosystem players to unlock new growth curves and break through growth bottlenecks in traditional businesses.

This article shares extensive insights on industry trends and customer pain points for AI-related service providers.

1. Industry development trend: The AI industry has gradually shifted from competition over foundational LLMs to competition over application layer deployment. AI search and AI agents are currently the fastest-growing application directions, with high recognition from industrial capital and enormous room for future growth.

2. Customer pain point insight: A large number of early-stage AI companies lack the capital and capacity to develop their own foundational LLMs, and urgently need asset-light solutions to enter the AI track. Users also have clear unmet demand for upgraded search experiences.

3. Business direction reference: Service providers can build supporting tools and solutions around multi-model routing and scheduling, and retrieval-augmented generation (RAG) development, to serve the large market of small and medium-sized AI startups. At the same time, the AI agent boom has driven sustained growth in demand for computing power, creating ongoing demand growth in fields including computing scheduling and supporting technical services. Service providers can join the open ecosystems of leading vendors to acquire stable customers.

Perplexity AI’s growth and NVIDIA’s ecosystem strategy offer multiple insights for AI platform operators around ecosystem building and operational management.

1. Understanding core demand from AI application companies: Most application-layer AI companies need stable supply of computing power and access to model resources. Companies relying on third-party models face clear supply chain risks, and platform providers can resolve this pain point for customers through ecosystem binding, while building their own competitive moats.

2. Referenceable platform layout model: NVIDIA’s "equity investment + ecosystem binding + chip sales" model is highly worth learning from. By investing in high-growth AI application companies, NVIDIA drives sales of its own chips and computing products, while building an open ecosystem to attract more players, forming a positive growth cycle.

3. Investment sourcing and risk mitigation: Companies in the AI search and AI agent tracks grow fast and have high demand for computing power, making them high-quality target for platform sourcing and should be prioritized for partnership. However, platforms need to be wary of valuation bubbles during布局, and focus on the real user demand and revenue growth of invested companies to avoid risks from false demand inflated by capital.

This article documents the latest industry developments in the AI search track, providing a vivid case study and clear research directions for industrial researchers.

1. New industry trend: High-growth independent leading players have already emerged at the AI application layer. Perplexity AI, founded less than four years ago, has crossed a $30 billion valuation to become the highest-valued independent player in the AI search track. AI industry competition has shifted from foundational LLMs to application layer deployment, and AI agents have become a new growth catalyst for the application layer.

2. New business model research: The asset-light "multi-model routing + retrieval-augmented generation" business model has been validated by the market and capital, breaking the industry perception that only companies building in-house foundational LLMs can succeed in the AI sector, and providing a new sample for researching growth paths of AI startups.

3. New questions worth researching: NVIDIA’s layout model of "equity investment tied to chip procurement" carries underlying risks. Whether it will fuel industry valuation bubbles, and how application-layer companies relying on third-party LLMs can resolve supply chain risks, are new questions emerging amid AI industry development that deserve in-depth research.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

这比上一轮融资后的200亿美元估值增长了超过50%。

硅谷头部AI公司的估值还在飙升。

据The Information报道,Perplexity AI正在完成新一轮股权融资,英伟达正在洽谈参与,此次规模预计达数十亿美元,完成后Perplexity的估值将超过300亿美元(约合2020亿元)。这比上一轮融资后的200亿美元估值增长了超过50%。

Perplexity AI是2022年成立的美国AI初创企业,创始人包括3名90后。2025年12月,足球明星C罗也宣布投资了Perplexity,并出任其全球品牌代言人。

这家公司最初主打AI对话搜索服务,目前已推出可自动执行电脑复杂任务的AI智能体产品。若本轮融资落地,这家成立仅四年的公司将成为AI搜索赛道上估值最高的独立玩家之一。

有意思的是,在讨论股权投资之前,英伟达最初考虑的是另一种方案:花数十亿美元购买Perplexity的部分技术授权,再挖走一些员工。但谈判最终转向了更传统的股权投资。

90后,把Google逼出舒适圈

很多人对Perplexity的印象还停留在“一个会搜索、会总结、更聪明的AI搜索”。你问一句,它直接给一段带来源链接的综合答案。如果只把它看成ChatGPT的搜索版,就会严重低估这家公司。

Perplexity的核心创始团队背景都带着顶尖AI实验室的标签。

灵魂人物是1994年的CEO Aravind Srinivas,来自印度,在印度理工学院马德拉斯分校完成本硕学习后,进入加州大学伯克利分校攻读计算机科学博士学位,先后在DeepMind、Google Brain实习,2021年回到OpenAI担任研究科学家。CTO Denis Yarats来自Meta AI,首席战略官Johnny Ho负责排名和后台系统,二位均是90后。

实际上,公司还有一位联创叫Andy Konwinski,是一名80后,由于他同时是Databricks的联合创始人,据悉没有全职参与。四个人凑在一起,最初的想法很简单:传统搜索引擎返回的是一堆链接,用户需要自己点进去筛选信息,为什么不直接给出提炼好的答案并附上出处?

这个思路在ChatGPT刚火的时候并不被看好。OpenAI在做通用大模型,Google有百万级用户的搜索入口,一个创业公司为何能插足?

Perplexity选了一条差异化路径:它不自己训练基础模型,而是根据不同任务调用和路由多家厂商的模型,再用实时联网检索做RAG(检索增强生成)。别人的模型它都能用,自己专注做检索和编排。轻资产,快迭代。

2022年12月,ChatGPT引爆全球生成式AI浪潮仅一个月后,诞生四个月的Perplexity就把产品形态做了出来。它能直接检索全网信息、提炼答案、附带来源链接。这一模式一度被视为“谷歌杀手”,把Google逼出了舒适圈。

这条路走通了,融资节奏也在加快。从2023年至2025年,估值就来到200亿美元。但真正让估值从200亿跳向300亿的催化剂,不是搜索,是一个叫Perplexity Computer的产品。

2026年2月25日,Perplexity推出了面向专业用户的云端AI智能体产品Perplexity Computer,能够自动执行电脑端的各类复杂工作任务。据报道,Perplexity Computer已与OpenAI、Anthropic的同类智能体产品形成直接竞争。

这款产品彻底改变了公司的增长曲线。2026年3月,Perplexity的年化经常性收入突破4.5亿美元,到8月,年化营收已进一步攀升至超过7.5亿美元,而2026年初这一数字尚不足2.5亿美元。不到八个月,营收翻了近两倍。

用户规模同样可观。Perplexity高管表示,其搜索与代理工具的月活跃用户已超过1亿,其中包括数万家企业客户。也有数据显示,月活跃用户在2026年已稳定在3000万至4500万之间。

从搜索到Agent,Perplexity完成了AI赛道最艰难的一次跨越。

英伟达的算盘

本轮估值300亿美元的关键投资方,是英伟达。

黄仁勋曾在采访里说过,他日常就用Perplexity。但CEO的个人偏好不足以解释一笔数十亿美元的投资。更底层的逻辑在于,Computer执行的任务越多、时间越长,消耗的算力也越多。Perplexity因此从一家搜索公司变成了算力大户,英伟达就能卖出更多芯片

双方的合作早已展开。Perplexity已表示将采用英伟达Vera中央处理器运行代理式AI查询,还与英伟达重点支持的云服务商CoreWeave签署了芯片使用协议。据The Information报道,双方代表目前每周多次会面,就硬件和软件开发展开沟通。

一手卖服务器,一手投应用入口,提前押注下一轮算力需求的制造者。这才是英伟达的底层逻辑。这个打法不是第一次了。

2024年英伟达就开始与OpenAI洽谈投资,2025年9月双方正式宣布战略合作,英伟达承诺投资最高1000亿美元用于OpenAI数据中心建设,OpenAI则承诺大量采购英伟达芯片。

2025年3月,微软宣布向AI初创公司Inflection AI支付约6.5亿美元获得其AI模型授权,并聘用其大部分员工;随后英伟达与微软共同参与了Anthropic的新一轮融资,其中英伟达承诺投资最高100亿美元用于加速计算基础设施,微软追加投资最高50亿美元,而Anthropic承诺向微软Azure采购数百亿美元的算力资源。这些交易的共同模式均是芯片供应与股权投资深度绑定。

据PitchBook统计,英伟达2025年参与了近67笔风投交易,高于2024年的54笔。这些被投公司里,不少转头就拿钱买回了英伟达的GPU。

今年3月,英伟达牵头成立了Nemotron联盟,Perplexity、Mistral、Cursor、Thinking Machines等一众明星团队被拉入同一阵营,共享算力与数据。

除洽谈投资Perplexity外,近日还传出英伟达已同意支付60亿美元获得AI编程初创公司Poolside的模型使用许可,吸纳约100名工程师,并以120亿美元投前估值向其额外投资10亿美元股权投资。这本质上也是变相人才收购,目标是加速自研Nemotron系列开源模型。对Poolside,英伟达想买的是技术和人;对Perplexity,英伟达最初也想这么干,但最终转向了股权投资。

客观来说,这套打法也有隐忧。英伟达投出去的钱,最终有多少能转化成真实的芯片采购,有多少只是估值泡沫?这些公司是真的有那么多用户需要算力,还是因为英伟达不断投钱,才让需求看起来特别旺?

A轮投资者回报达到200倍

Perplexity的融资速度,属于典型的资本神话。

2022年9月,在产品尚未上线时,Perplexity就获得了310万美元的种子轮投资。2023年3月,公司完成2560万美元的A轮融资,估值达到1.5亿美元。

进入2024年:1月,公司完成由IVP领投的B轮融资,融资金额为7360万美元,估值达到5.2亿美元;4月,Perplexity宣布完成约6300万美元融资,估值超过10亿美元,正式跻身独角兽行列;几个月后,Perplexity又完成2.5亿美元的新一轮融资,软银参投,估值达到30亿美元。

2025年,公司完成3轮融资,使公司估值飙升至200亿美元,足球明星C罗也在此时进场。

Perplexity的投资人名单堪称豪华:英伟达、软银、Databricks等产业资本齐聚一堂;杰夫·贝佐斯、前推特副总裁Elad Gil、前GitHub CEO Nat Friedman等科技界大佬个人下注;甚至连谷歌AI掌门人Jeff Dean和YouTube前CEO Susan Wojcicki也赫然在列。这种来自竞争对手高管的投资,足以说明其含金量。

到2026年初,又一轮将估值推至约212亿美元,投资方未披露。本次新一轮融资再度加码到300亿美元,不到四年,翻了200倍。

在回报上,以A轮投资人进入时的估值1.5亿美元计,三年半回报200倍。就算最后一轮进来的投资人(2026年初212亿估值),到如今300亿,账面也浮盈40%。

Perplexity年化收入为7.5亿美元,300亿美元估值意味着P/S约40倍。可以横向对比的是:OpenAI最新估值约5000亿美元,年化收入约200亿美元,P/S约34倍;Anthropic估值约850亿美元,年化收入约50亿美元,P/S约20.5倍。Perplexity的收入增速确实更猛,8个月翻了三倍。但40倍P/S,只有收入继续指数级增长才能扛得住。

但实际上,Perplexity并不自研基础大模型,其Sonar模型基于Meta的Llama微调,核心能力建立在多模型路由和检索增强之上。其最强能力的答案依赖OpenAI、Anthropic、Google和xAI的第三方模型,这些模型一旦涨价或限制访问,Perplexity的成本结构将承压。它和这些“供应商”之间,既是客户,又是对手。

好在,黄仁勋的出现,恰好补上了这块拼图。英伟达正在用Nemotron联盟打造自己的开源模型生态,Perplexity是这个联盟的核心节点之一。如果Perplexity未来能更多地依赖英伟达系模型而非竞争对手,300亿美元的估值就有了更坚实的底盘。这笔投资,既是财务下注,也是生态绑定。

Perplexity CEO Srinivas此前公开表示,公司计划于2028年推动IPO。也就是说,这轮300亿美元估值的融资,大概率是上市前的倒数第二轮甚至最后一轮。

注:文/黎曼,文章来源:投中网(公众号ID:China-Venture),本文为作者独立观点,不代表亿邦动力立场。

文章来源:投中网

广告
微信
朋友圈

FAQ回顾

Perplexity AI是什么公司?

Perplexity AI是2022年成立的美国AI初创企业,核心创始团队多为90后,最初主打AI对话搜索服务,目前已推出可自动执行电脑复杂任务的AI智能体产品Perplexity Computer,是AI搜索赛道估值最高的独立玩家之一。

英伟达为什么要投资Perplexity AI?

英伟达投资Perplexity核心逻辑在于其AI智能体产品消耗算力多,可拉动英伟达芯片销售;同时将其纳入Nemotron联盟核心节点绑定生态,提前押注下一轮算力需求制造者,属于芯片供应与股权投资深度绑定的常规打法。

Perplexity AI的经营数据表现如何?

2026年8月Perplexity年化营收已超7.5亿美元,不到8个月翻近两倍;其搜索与代理工具月活跃用户超1亿,包含数万家企业客户,另有数据显示2026年其月活稳定在3000万至4500万之间。

Perplexity AI的核心发展路径是什么?

Perplexity走差异化发展路径,不自研基础大模型,而是根据不同任务调用和路由多家厂商的模型,用实时联网检索做检索增强生成,自身专注做检索和编排,具备轻资产、快迭代的发展优势。

Perplexity AI有上市计划吗?

Perplexity CEO Srinivas公开表示,公司计划于2028年推动IPO,本轮300亿美元估值的融资大概率是上市前的倒数第二轮甚至最后一轮融资。

这么好看,分享一下?

朋友圈 分享

APP内打开

+1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0