AI-powered search engine Perplexity AI, Valued at $520M, Secures $70M in Funding

TL;DR:

  • Perplexity AI, valued at $520 million, successfully raises $70 million in funding.
  • The funding round, led by IVP, includes investments from NEA, Databricks Ventures, Elad Gil, Tobi Lutke, Nat Friedman, and Guillermo Rauch, among others.
  • Perplexity AI, operational since August 2022, is challenging established search engine giants with its chatbot-like interface and source-citing capabilities.
  • The platform offers a Pro plan with multiple GenAI models and advanced features for $20 per month.
  • Concerns arise over the sustainability of GenAI search tools due to high operational costs and potential misuse.
  • Perplexity’s annual recurring revenue is estimated at $5-10 million, but substantial expenses for GenAI model training are a consideration.
  • Copyright issues and potential anticompetitive impacts on publishers are additional challenges for GenAI search tools.
  • Despite uncertainties, investors remain confident, and Perplexity AI continues to expand.

Main AI News:

In the ever-evolving landscape of AI-powered search engines, Perplexity AI stands out as a rising star, determined to redefine the industry’s standards. While facing formidable giants like Google, these innovative startups are driven to create a niche by offering an unparalleled user experience.

Today, Perplexity AI proudly announced its successful fundraising round, securing $70 million in capital. This impressive achievement was made possible through investments from prominent firms like IVP, NEA, Databricks Ventures, as well as notable individuals such as Elad Gil, Tobi Lutke, Nat Friedman, and Guillermo Rauch. Notably, the round also included participation from Nvidia and the prominent figure, Jeff Bezos.

According to sources close to the matter, this fundraising round has elevated Perplexity’s post-money valuation to $520 million. While this may appear modest when compared to the colossal valuations of other GenAI startups, it’s worth acknowledging that Perplexity has only been operational since August 2022, making this ascent truly remarkable.

Founded by a team of experts, including Aravind Srinivas, Denis Yarats, Johnny Ho, and Andy Konwinski, Perplexity AI boasts a roster of engineers with extensive backgrounds in AI, distributed systems, search engines, and databases. Srinivas, the CEO, previously contributed to OpenAI’s research endeavors, focusing on language and GenAI models such as Stable Diffusion and DALL-E 3.

Setting itself apart from conventional search engines, Perplexity AI offers a chatbot-like interface that empowers users to pose questions in natural language. Whether inquiring about topics like calorie burning during sleep or the least-visited countries, Perplexity’s AI provides concise responses with source citations from websites and articles. This feature allows users to delve deeper into their chosen subjects through follow-up questions.

With Perplexity, users can obtain instant answers to any query, complete with sources and citations,” Srinivas affirmed. “Perplexity caters to anyone and everyone who relies on technology for information retrieval.”

At the core of the Perplexity platform lies a collection of GenAI models, both developed in-house and by third-party sources. Subscribers to Perplexity’s Pro plan, priced at $20 per month, gain access to an array of models, including Google’s Gemini, Mistra 7Bl, Anthropic’s Claude 2.1, and OpenAI’s GPT-4. This allows users to enjoy features like image generation, unlimited use of Perplexity’s Copilot, and the ability to upload documents, including images, for model analysis.

It’s worth noting that Perplexity’s offerings bear similarities to competitors such as Google’s Bard, Microsoft’s Copilot, and ChatGPT, with its chat-centric user interface reflecting the current trend in GenAI tools.

In addition to established rivals, the startup You.com also provides AI-powered summarization and source citation tools, optionally powered by GPT-4.

Srinivas contends that Perplexity distinguishes itself by offering robust search filtering and discovery options. Users can tailor their searches to academic papers or explore trending topics submitted by fellow platform users. While these differentiators are impressive, some skeptics question whether they can withstand replication by others in the industry.

Beyond its core search function, Perplexity is expanding into serving its own GenAI models. Leveraging the platform’s search index and the broader web, these models aim to enhance performance and are available through an API exclusively for Pro customers.

However, concerns persist regarding the sustainability of GenAI search tools, primarily due to the considerable costs associated with running AI models. For instance, OpenAI spent approximately $700,000 daily to support ChatGPT’s demand, while Microsoft reportedly incurs an average monthly loss of $20 per user on its AI code generator.

Sources familiar with the matter reveal that Perplexity’s annual recurring revenue currently ranges between $5 million and $10 million. While seemingly healthy, this revenue must be weighed against the substantial expenses involved in training and maintaining GenAI models like those used by Perplexity.

Furthermore, the potential for misuse and the spread of misinformation remains a pressing issue with GenAI search tools. AI models, while powerful, are not infallible and may miss critical details, misinterpret language, or present inaccurate information with unwavering authority. They are also susceptible to biases and toxic content, as demonstrated by Perplexity’s own models.

Another hurdle on Perplexity’s journey to success pertains to copyright concerns. GenAI models rely on learning from examples to create content, often scraping millions to billions of examples from the web for training datasets. While vendors argue that the fair use doctrine protects their web-scraping practices, copyright holders, including artists and authors, have contested this and initiated lawsuits seeking compensation.

Interestingly, Perplexity differs from some of its peers by not offering IP protection to its customers. As outlined in its terms of service, customers must “hold harmless” Perplexity from claims, damages, and liabilities arising from their use of the service, effectively exempting Perplexity from legal fees in potential copyright disputes.

Moreover, GenAI search tools have faced criticism for potentially diverting traffic, content, and ad revenue from publishers through anticompetitive means. The rise of AI in search, as evidenced by a model from The Atlantic, suggests that search engines integrated with AI can answer users’ queries without requiring them to navigate to external websites. While some vendors, like OpenAI, have entered agreements with select news publishers, many, including Perplexity, have not.

Srinivas views this trend as a feature rather than a flaw. He emphasizes that Perplexity eliminates the need to sift through SEO spam, sponsored links, and multiple sources, offering a more efficient approach to knowledge acquisition and sharing. This, he believes, will usher society into an era of accelerated learning and research.

Despite the numerous uncertainties surrounding Perplexity’s business model and the broader landscape of GenAI and consumer search, investors remain undeterred. With an active user base of 10 million monthly users, Perplexity has secured over $100 million in funding to date. This substantial capital injection will be directed toward expanding the company’s 39-person team and enhancing product functionality, according to Srinivas.

Conclusion:

Perplexity AI’s successful funding round reflects growing interest in AI-powered search tools. While challenges related to costs, copyright, and competition persist, the potential for revolutionizing knowledge acquisition is significant. The market for GenAI search tools is poised for growth, with startups like Perplexity AI driving innovation and attracting substantial investments.

Source