Nvidia Plans to Lead Investment in Perplexity and Take Stakes in Data Center Power Companies
Nashnova编辑部
Nvidia is preparing to lead a new funding round for AI search company Perplexity at a valuation above $30 billion, while also taking a minority stake in data-center infrastructure firm Cloverleaf. Both moves point to the same shift: Nvidia is evolving from chip supplier to the capital backbone of the entire AI stack.
Is Perplexity worth $30 billion?
The new round values Perplexity above $30 billion — up more than 50% from the prior round about a year ago.
Annualized revenue has jumped from under $250 million in early 2026 to over $750 million, driven largely by Perplexity Computer — an AI agent that automates desktop tasks for professional users.
This means → the revenue numbers can justify the price tag, but the real test is the next question — traffic.
Revenue is surging — so why is traffic falling?
Wix's analysis of AI search traffic shows Perplexity's volume dropped 19.7% in Q1 2026, while Google's Gemini and Anthropic's Claude both grew.
StatCounter data also shows Perplexity's global AI chatbot referral share slipping in August 2026 — though these metrics don't capture in-app usage or products like Comet.
In plain terms = users may not have left, but fewer are arriving through web-search entry points. Rivals — OpenAI, Google, Anthropic — are all adding search and agent features to their own products, pulling traffic away.
Why is Nvidia investing in Perplexity?
Nvidia isn't just writing a check — it's using equity to lock a leading AI application into its hardware ecosystem.
This means → Perplexity runs inference on Nvidia chips; Nvidia uses equity to secure that demand layer. Hardware and software feed each other.
CEO Aravind Srinivas has said the company is considering an IPO around 2028. Nvidia's entry now is also a bet on pre-IPO valuation growth.
What's the logic behind buying into a power company?
Nvidia acquired a minority stake in Cloverleaf Infrastructure, a data-center infrastructure firm. This is the third equity investment Nvidia has made recently in U.S. data-center land and power.
In plain terms = chips need somewhere to plug in and run. Nvidia is locking down land and power ahead of competitors — because selling a chip that the buyer can't deploy is the same as not selling it.
This reflects a shifting bottleneck: AI compute is moving from "can't buy chips" to "have chips but no power, no facility."
How big is Nvidia's capital footprint now?
Regulatory filings show Nvidia invested $18.6 billion in private companies and infrastructure funds in fiscal Q1 2027; its unlisted equity portfolio stood at roughly $42.3 billion as of April.
Last week Nvidia paid $6 billion for a non-exclusive technology license from AI model-development software firm Poolside, invested $1 billion at a $12 billion valuation, and offered positions to over 100 employees who worked on Poolside's Laguna model. It has also committed more than $100 billion in credit support for OpenAI's Ohio data-center project.
This means → Nvidia is no longer just a chip company. It is deploying tens of billions of dollars to tie every link of AI infrastructure — software, models, power, land — to itself.
Can this valuation be justified?
The core tension: revenue is growing fast, but traffic share is falling — two signals pulling in opposite directions.
If Perplexity's agent product (Perplexity Computer) can unlock new paid use cases, revenue growth can stand independent of search traffic.
In plain terms = the $30 billion valuation is not a bet on "AI search." It's a bet on the bigger story of "AI agents" — and whether that story holds depends on how many users are willing to pay.
Content is for reference only, not financial advice.