Listen to this story
OpenAI’s enviable success is driving startups to experiment with Large Language Models (LLM) and build their own text generation models. A large portion of the current AI market is driven by companies making generative and LLM-based models, which has eventually led to VCs looking to pump in a huge chunk of their funds into the field.
With the AI excitement back in investors, most are eager to invest in teams and startups looking to solve problems with LLMs. It looks like the only question VCs ask these days is whether the AI startup is working with LLMs or not. If the answer is affirmative, the money pours in.
The question is whether this huge interest in these specific technological startups is because the investors are actually hoping for AI to grow further, or are they hopping onto the bandwagon purely for financial gains.
Subscribe to our Newsletter
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
After speaking with many VCs investing in AI startups, we found that most of them genuinely find the generative AI field interesting. But how much of it is because of staying relevant in the market versus actually understanding the technology, is an unexplored question.
Most investors believe that Indian startups mostly work on application-based AI, and not building the infrastructure for it. Though this sounds like a critique of startups, that is what investors are actually looking for. Som Pal Choudhury, partner at Bharat Innovation Fund, said running large scale models is extremely resource intensive. “We believe that a plethora of chatbot companies will simply integrate ChatGPT as the underlying engine,” said Choudhury. The firm is closely monitoring the space and said that multiple companies have already approached them that have these open-source models running underneath their workings.
Similar views were expressed by Ajay Jain, co-founder of Silverneedle Ventures. He said that ChatGPT clearly shows an inflexion point in the entire natural language processing space and we are seeing something tangible. He further added that this is the first of a kind application where we are talking about creating content with specific use-cases that are wrapped into solutions/products that can grow exponentially. “One could witness some unicorns that work in the generative AI space to produce moots on how it solves daily user problems with little care about how the magic is happening,” said Jain.
That is definitely true. Though Google, Meta, and Microsoft are leading the space, there are various startups being increasingly funded by various investors. Jasper, an AI-powered copywriting tool, raised $125 million and became a unicorn with a valuation of $1.5 billion. Similar was the case with Anthropic, founded by former VP of research of OpenAI, Dario Amodei, that raised $580 million in April and is working on building LLMs similar to GPT-3. Cohera, a developer-focused NLP toolkit, also raised $125 million. Another text-generation startup for automating blog posts and emails, Regie.ai, raised $10 million in Series A from Scale Venture Partners.
Navigating the Tech-Dabbling Deliberation
Most investors rely on technology experts, often outsourcing their services to assess the startups. Though VCs might have a sound technological background, the understanding of money and financial returns is much more than literacy about AI.
The case isn’t as bad as it sounds because ultimately the AI field is getting flooded with AI and large language-based startups, resulting in democratisation of the field by taking the control away from the big-techs. Sonya Huang, a partner at Sequoia Capital, wrote a blog post where she described the various business opportunities in generative AI, that includes LLMs. She wrote that generative AI tools can “generate trillions of dollars of economic value”.
Economic value is, obviously, the key determinant for VCs. Now, a majority of the content on social media is about generative models like ChatGPT, DALL.E, or Stable Diffusion. The field that was seldom spoken about till 2021, is suddenly gaining traction among the VC circles. “My peer group of AI-focused investors have never been as busy,” said Sandhya Hegde, partner at Unusual Ventures. Gaurav Gupta from Lightspeed Ventures echoed similar sentiments and said that technologies like these can benefit every field, and should definitely be funded more.
A lot of valuations got corrected with the onset of the apparent ‘funding-winter’. These views were expressed by all the investors that AIM spoke with. This has resulted in VCs jumping onto the wagon early like seed or pre-seed, for being part of the expansive field. A lot of VCs also believe that making generative models is very easy at the moment as big-tech companies like Google and Microsoft have already built foundation models – all that the startups have to do is to mould them for specific use cases.
Most of the VCs are following the footsteps of the legends like Khosla Ventures, who were the first ones to invest in OpenAI with Microsoft now continuously pouring funds into it. Firms like Sequoia Capital and Y Combinator have been heavily investing in generative AI startups and now every firm wants to be known as Khosla Ventures, Sequoia, or Microsoft of OpenAI.