AI-powered search engine Perplexity AI, now valued at $520M, raises $70M

Search engine

Perplexity AI, valued at $520 million, has secured a $70 million investment in a funding round that signals a new wave of competition in the AI-powered search engine arena. While juggernauts like Google fortify their platforms with cutting-edge AI technology, emerging startups aim to revolutionize the domain of AI-driven search from its core. Surmounting the challenge of competing against established players with colossal user bases may seem daunting, yet these up-and-comers are confident they can carve out a niche by offering an exceptional user experience.

Among these ambitious entrants, Perplexity AI disclosed its recent funding round, spearheaded by IVP and supported by investments from NEA, Databricks Ventures, industry luminaries like Elad Gil, Tobi Lutke (CEO of Shopify), Nat Friedman (former GitHub CEO), and Guillermo Rauch (Vercel founder). Notably, Jeff Bezos also participated in the investment, propelling Perplexity’s post-money valuation to $520 million.

Although this valuation might seem modest compared to other gen AI startups, considering Perplexity’s inception in August 2022, this ascent is indeed commendable. Founded by Aravind Srinivas, Denis Yarats, Johnny Ho, and Andy Konwinski, a team comprising experts in AI, distributed systems, search engines, and databases, Perplexity distinguishes itself by offering a chatbot-like interface. Users can engage with the platform using natural language queries, receiving comprehensive responses with source citations. This unique approach enables users to delve deeper into various subjects through follow-up questions, setting Perplexity apart from traditional search engines.

“Perplexity provides instant and well-sourced answers to any inquiry,” Srinivas explained. “It caters to tech users across the spectrum, offering a comprehensive solution for information searches.”

At the core of Perplexity’s platform lies an ensemble of in-house and third-party gen AI models. Subscribers to the Pro plan ($20 per month) gain access to a range of models, including Google’s Gemini, Mistra 7Bl, Anthropic’s Claude 2.1, and OpenAI’s GPT-4, allowing them to toggle between these models. This subscription unlocks various features such as image generation, unlimited use of Perplexity’s Copilot feature – which tailors searches based on individual preferences – and file uploads. The file upload functionality permits users to upload documents, including images, for analysis by the models to extract specific information (e.g., “Summarize pages 2 and 4”).

If this user experience sounds akin to Google’s Bard, Microsoft’s Copilot, and ChatGPT, the similarity is evident. Moreover, Perplexity’s chat-oriented user interface echoes the style of today’s most popular gen AI tools.

Apart from these direct competitors, the search engine startup You.com offers analogous AI-driven summarization and source-citing features, optionally powered by GPT-4.

Srinivas argues that Perplexity offers more extensive search filtering and discovery options than most competitors, such as enabling users to narrow searches to academic papers or explore trending search topics contributed by other users. While these features seem distinctive, I’m not entirely convinced that they are so unique that they couldn’t be replicated — or haven’t already been replicated, for that matter. However, Perplexity is broadening its scope beyond search by introducing its gen AI models that utilize the platform’s search index and the public web. These models are accessible through an API for Pro customers, presumably aiming to enhance performance.

There’s skepticism about the sustainability of gen AI search tools for various reasons, primarily due to the high operational costs of AI models. For instance, OpenAI incurred expenses of around $700,000 daily to sustain ChatGPT. Microsoft reportedly faces an average loss of $20 per user monthly on its AI code generator.

Insiders familiar with the situation informed TechCrunch that Perplexity’s yearly recurring revenue currently stands between $5 million and $10 million, a seemingly healthy figure. However, this seems less robust when considering the substantial expenses involved in training gen AI models like those of Perplexity.

Concerns about the potential misuse and dissemination of misinformation through gen AI search tools like Perplexity are valid. AI doesn’t consistently summarize information accurately, occasionally missing crucial details, distorting language, or presenting incorrect facts confidently. Moreover, it’s prone to exhibiting bias and generating toxic content, as recently demonstrated by Perplexity’s own models.

Another obstacle for Perplexity’s success is the issue of copyright. Gen AI models learn from various examples across the web to generate content like essays, code, emails, and articles. Many vendors, including Perplexity, presumably gather examples from the web to enrich their training datasets. While vendors argue that their web-scraping practices are protected by the fair use doctrine, copyright holders like artists and authors disagree, resulting in lawsuits seeking compensation.

On a related note, while many gen AI providers offer protections shielding customers from intellectual property claims, Perplexity doesn’t extend this safeguard. As per the company’s terms of service, users agree to indemnify Perplexity from claims, damages, and liabilities arising from the use of its services — absolving Perplexity from legal responsibilities.

Several plaintiffs, like The New York Times, argue that gen AI search platforms drain publishers’ content, audience, and advertising revenue through potentially anticompetitive practices. Despite debates on the anticompetitive nature, this technology undeniably impacts web traffic. A study by The Atlantic suggests that if an AI-equipped search engine like Google were integrated, it would directly answer a user’s query without requiring a visit to the website 75% of the time. While some vendors, such as OpenAI, have struck deals with specific news publishers, most, including Perplexity, haven’t followed suit.

Srinivas frames this as a positive aspect — not a flaw. He highlights, “[With Perplexity, there’s] no need to click on different links, compare answers, or endlessly dig for information.” He envisions an era without the hassle of sifting through SEO clutter and sponsored links, heralding a more efficient way of acquiring and disseminating knowledge, ultimately propelling society into an age of accelerated learning and research.

Despite the uncertainties surrounding Perplexity’s business model and the broader landscape of gen AI and consumer search, investors remain undeterred. The startup, boasting ten million active monthly users, has amassed over $100 million in funding to date. A substantial portion is allocated to expanding its team of 39 employees and enhancing new product functionalities, as per Srinivas.

“Perplexity is ambitiously crafting a product capable of democratizing the power of AI for billions,” remarked Cack Wilhelm, a general partner at IVP, emphasizing Aravind’s ability to uphold a far-reaching, long-term vision while consistently delivering product advancements, essential qualities to tackle the pivotal and fundamental problem of search.

Leave a Reply

Your email address will not be published. Required fields are marked *

Scroll to Top