I’ve been covering generative AI in VC Cafe for the past couple of years. In November I wrote that generative AI will become truly mainstream when it moves from playful to useful. We seem to have hit that inflection point in January, merely two months after the public launch of ChatGPT when the service hit 100 million users, making it one of the fastest growing consumer products of all time. Now that OpenAI launched a ChatGPT API (using its GPT 3.5 model) we expect to see the next evolution of the application layer of Generative AI services powered by this technology.
The buzz around generative AI has fuelled a new wave of VC investments, even in the current market, as the number and volume of venture backed deals dropped significantly since the second half of 2022. According to The Economist, Generative AI startups have raised $11 billion to date, not counting the $10 billion Microsoft investment into OpenAI (see chart below).
Big tech is piling in on the Gen AI action
The past month has also brought several announcements around generative AI from big tech. Microsoft moved fast and started incorporating Generative AI tools in its search (Bing) and MS Office suite. Google announced it is working on BARD, its own version of an AI powered chatbot that will be incorporated in search and the Gsuite tools. YouTube’s new CEO, Neal Mohan, said the company is embracing generative AI and will start offering tools to its creators. Meta created a team devoted to generative AI and announced its own large language model (LLM) called LLaMa, trained on publicly available datasets only (thus removing some of the copyright concerns around ChatGPT). Meta also shared it is working on ‘AI personas’ for Whatsapp, Messenger and Instagram as well as Generative AI features for its VR products. Amazon partnered with Hugging Face to work on Bloom, an open source multilingual LLM and Snap announced it will incorporate Generative AI features in its Snap+ product. Quite a month!
Of course it’s not just the big companies who are active in the generative AI space. Just as an example, there are over 40 generative AI startups in the current Ycombinator batch (YC W23) alone!
What are the leading venture investors saying about generative AI?
In light of the rapid advancement of generative AI, many investors are eager to take advantage of this technology’s potential for disruptive innovation and lucrative returns. If you ask me, generative AI is not a blip or hype – it truly represents a jump in capabilities and AI will be embedded in every product as the technology improves and costs come down. In the early days of the iPhone (2007-2010), the apps were playful and perhaps a bit of a novelty, but within less than 10 years we got Uber, Waze, Shazam, Snap, Whatsapp, etc. Utilities which quickly became part of our everyday lives. I think Generative AI will follow a similar path.
Here’s what some of Silicon Valley’s digerati said about generative AI in the past few weeks:
Marc Andreessen, A16Z “It’s going to hit hard and change the way we do things”
Doug Leone, former Managing Partner at Sequoia “AI is real. AI is the next major platform” (from Invest like the best podcast)
Pete Flint, NFX: Generative AI is going to change everything
David Sachs, Craft Ventures: “AI is the autocomplete for everything in enterprise”
Sandhya Venkatachalam, partner at Khosla Ventures and early investor in OpenAI compared recent advancements in generative AI to the creation of the internet itself. “I think this is absolutely on the same order of magnitude. That’s a personal belief.”
As AI researcher and entrepreneur Ross Goodwin puts it, “Generative models could transform every single media industry on the planet. I don’t think that’s an overstatement.“
Ok, but we’ve seen hype bubbles before. Isn’t this just like the Crypto or Web3 craze?
For the past two decades, Silicon Valley has lacked a true technological breakthrough. In the ’80s, we had the advent of the personal computer; in the ’90s, the internet; and in the 2000s, the mobile phone and the suite of apps built on it. Since then, the tech world has been waiting for the next big invention (some are still bullish it could be Web3 or AR/VR). Now, many are seeing generative AI as a contender.
Shirin Ghaffary, Vox
Generative AI is different from web3 or crypto hype. In contrast to web3 and crypto, generative AI promises to fundamentally change the way we create and consume content. For instance, generative AI can be used to create original music, visuals, and even to write or translate text with no human input. In the context of the enterprise, generative AI can be used to automate customer support, write code and support designers and product managers in their roles using technology.
It can also create new markets. A prime example of the potential of generative AI is the creation of deepfakes or synthetic video, which use AI algorithms to create vide from text prompts, using real human characters (digital twins). While deepfakes have raised concerns about their potential use in spreading misinformation or committing fraud, they also showcase the incredible power of generative AI to manipulate media in ways that were previously impossible. As the costs of creating high quality videos goes down, they can be used more broadly for education, training, communication etc.
For example, below is a video I created on the role of AI in education using Hour One in five minutes. ChatGPT helped me write the script, and HourOne’s platform did the rest (disclosure: HourOne is a Remagine Ventures portfolio company).
It’s important to be aware of Generative AI’s limitations and risks
As I mentioned in my post “Investing in Generative AI tools“, investors have a tough job here sifting the wheat from the chaff. The foundation models like OpenAI’s GPT, AI21 Labs Jurassic, or Stability AI’s open source model for image creation, or Google-funded Anthropic, are expensive to run and fund. Also, the training of generative models requires vast amounts of high-quality data, which can be difficult and expensive to obtain (unless you’re a large company).
In addition, IP concerns make investing in this space a legal minefield. On the one hand, companies like Stability and MidJourney are being sued for scraping copyright protected material from the Internet to train the models. On the other hand, the court already ruled in a case that the AI generated works cannot be copyright protected.
There are also ethical concerns associated with investing in generative AI beyond its technical limitations. It can be difficult to distinguish authentic content from fabricated content generated by generative models. The implications of this include fake news, cybercrime, and privacy violations.
We’re in the peak of the hype cycle for generative AI and it’s safe to assume that many of the companies that are being funded in this space will fail. So was the case with the early Internet, mobile, VR, etc. However, I believe that similar to cloud, AI is something we will expect to have in products going forward. And companies that fail to lean in and explore what are the implications of these technologies on their products or industries might be left behind. At a minimum, they will be more inefficient than their AI-powered competitors.
It’s without a doubt an interesting time to be building in this space and we’re excited about the potential of generative AI technology to enrich, empower and personalise content creation and innovation in health, education, commerce, entertainment, gaming and more.
As early stage investors, we’re actively reviewing many interesting opportunities in this space. We’re excited to meet with ambitious founders looking for compelling uses cases and will give any project a fair look and feedback.