Summary of Key Points
Perplexity AI, a young American AI startup founded just four years ago, is in the process of completing a new round of financing (with Nvidia participating), and its valuation is set to exceed $30 billion (a 50% increase from the previous round’s $20 billion). Founded by a team of post-90s entrepreneurs, the company started with AI search and has achieved explosive revenue growth through its “agents” – intelligent systems capable of automatically performing complex computer tasks. Its annual revenue has tripled in just eight months, reaching $750 million, and the user base has also grown rapidly. Nvidia’s investment strategy is to align with the demand for computing power (agents consume a large number of chips). Although Perplexity has created a capital myth (with its valuation increasing 200 times in four years), it also faces the risk of being dependent on third-party large models. The company plans to go public in 2028.
1. A 50% Valuation Jump: Agent Products as the “Growth Engine”
The key to Perplexity’s valuation increase from $20 billion to $30 billion is not its original AI search technology, but rather the Perplexity Computer – an AI agent that can automatically complete computer tasks, such as organizing data, writing reports, and even operating software. This product has completely transformed the company’s growth trajectory: its annual revenue was less than $250 million at the beginning of 2026 but soared to $750 million by August (tripling in just eight months), with over 100 million monthly active users, including tens of thousands of corporate clients. Previously, it was just a “smarter search tool” that provided direct links to answers; now, it has become a capable assistant, which is what has attracted higher valuations from investors.
2. Nvidia’s Investment: A Business of Selling Chips
Why is Nvidia investing billions in Perplexity? It’s not because Jensen Huang personally likes the company’s products, but because the more agents are used, the more computing power is consumed (AI models require numerous GPU chips). Nvidia’s approach is clear: it first invests in AI companies and then encourages them to purchase its chips. For example, it has invested in OpenAI and Anthropic, tying these investments to chip purchases. With Perplexity, it initially wanted to acquire the technology and talent directly but later opted for equity investment, which is more cost-effective and ensures a long-term partnership. Additionally, Nvidia has brought Perplexity into the “Nemotron Alliance” to share computing power and data, integrating it into its ecosystem to secure future chip sales.
3. The Post-90s Team: Differentiation That Competes with Google
Perplexity’s founding team consists of top talents in the AI industry: the CEO is a 1994-born Indian engineer with a Ph.D. from Berkeley and previous internships at DeepMind and OpenAI; the CTO comes from Meta AI, and both are post-90s. Their strategy is to avoid developing their own large models (saving money and time) by leveraging existing models from companies like OpenAI and Anthropic, combined with real-time online searches to provide users with well-organized answers along with source links. This approach is more convenient than Google’s traditional search model, which merely provides links for users to find information on their own, and has been dubbed a “Google killer,” forcing Google to launch similar products. Perplexity’s rapid iteration and low-cost model also allow them to respond quickly to market trends (for example, they launched their product just a month after ChatGPT became popular).
4. The Sweetness and Challenges of the Capital Myth
The Good News: The company’s fundraising has been incredibly fast – a $3.1 million seed round in 2022, a $150 million Series A round in 2023, and a $20 billion valuation in 2025, representing a 200-fold increase in four years (with Series A investors earning a 200-fold return). The investor lineup is impressive, including SoftBank, Jeff Bezos, Google’s AI leadership, and Cristiano Ronaldo. The Challenges: The valuation of $30 billion is 40 times its annual revenue, which is higher than OpenAI’s (34 times) and Anthropic’s (20.5 times), indicating a need for sustained exponential growth to maintain this ratio. Moreover, Perplexity’s dependence on third-party models (such as OpenAI) means that cost increases could arise if these models become more expensive or their access is restricted.
5. The Future: The Final Sprint Before Listing?
Perplexity plans to go public in 2028, and this round of financing is likely the second-to-last (or final) round before listing. Its partnership with Nvidia can help reduce its reliance on third-party models by using Nvidia’s open-source models, potentially stabilizing its valuation. However, whether it can maintain a valuation of 40 times its revenue will depend on the growth of its agent business. If corporate clients continue to purchase its services and revenue continues to rise, the company may have a successful listing; otherwise, the valuation could bubble. Overall, Perplexity is a “dark horse” in the AI industry, but its high valuation reflects both real growth and speculative capital interest. Nvidia’s investment is a win-win situation: it earns equity gains while also selling more chips. Perplexity’s future success will hinge on its ability to reduce its dependence on third-party models and continue to dominate the market with its agent products.