Dear Reader,
There’s no better place to “see” the early stages of a bull market in a specific technology sector than to watch what’s happening in the private markets.
Venture capitalists (VCs), private equity firms, and angel investors all have an inside view on what’s actually being built, how quickly the technology is advancing, and of course the early signs of adoption rates.
These factors are key inputs in determining whether or not to invest in an individual company or sector, and at what valuation. When things are normal, valuations for VC rounds tend to fit in a certain range appropriate for the round and where the company is in terms of product development, revenue generation, and path to free cash flow.
But when a sector is hot, most of those metrics are thrown out the window. No matter how much analysis is done to determine a fair valuation, it’s the capital supply vs. demand dynamic that ends up driving valuation ranges much higher. This is easy to see in the chart below.
The above chart shows the median venture capital pre-money valuation for generative artificial intelligence (AI) deals. The pre-money valuation is the valuation at which a VC firm or angel investor invests. After years of median valuations for generative AI startups in the $5-12 million range, 2021 and 2022 jumped remarkably to around $40 million.
This was largely driven by the low interest rate policy and pandemic-related stimulus in 2021 and into 2022. But despite the record pace of interest rate hikes over the last 15 months and the general decline in venture capital investments, the generative AI sector is booming. This will come as no surprise to Bleeding Edge subscribers. In the first quarter of this year alone, the median valuation has popped to $90 million.
The reality is that if the average were used instead of the median, the number would be far higher. Case in point, Microsoft plopped another $10 billion down for OpenAI at a reported $29 billion valuation. There have been a few other $1 billion-plus valuations as well.
The last time we saw this kind of excitement around a sector in the private markets was in the blockchain industry. But the differences between the two sectors are large. Many of the blockchain deals were for tokens and not equity. And as we know, the regulatory environment has been uncertain at best and antagonistic at worst.
Generative AI is different. Equity is on offer, and the path toward monetization is clear. Better yet, these generative AI companies are acquisition targets right out of the gate, and the best of them have an obvious path toward an IPO years down the road.
And that’s why the VCs are paying so much more – in terms of valuation – just to get a piece of a generative AI deal. The competition is fierce right now. Everyone wants exposure, which means that there is too much capital chasing too little allocation. That’s why the median valuation has spiked.
While we’ve been having a lot of fun sharing all of the emerging applications for generative AI in The Bleeding Edge over the last several months, most developments are still in the beta stage or early deployment. These examples are always useful signposts of what’s to come. But nothing beats solid numbers like the chart above.
A 9X increase in valuation since 2020 and a corresponding spike in invested capital drive crazy advancements in this technology, as has already been evidenced in The Bleeding Edge. And the adoption numbers are equally impressive.
Microsoft just reported 500 million chats through its own implementation of ChatGPT in its Bing Chat feature in the first 90 days! It has more than 100 million daily active users for Bing right now and growing. Those users have created more than 200 million images with the Bing image creator, based on OpenAI’s DALL-E text-to-image generator. And daily installs of the Bing mobile application have jumped 4X since the product launch.
With numbers like these, we know something incredible is happening. Generative AI is a once-in-a-generation technology, as is its investment opportunity.
One of the most interesting competitive dynamics that I’ve ever seen in my career has been watching how the rise of generative artificial intelligence (AI) caught Google off-guard. This has been particularly ironic considering that the seminal paper on generative AI was actually written by seven Google employees back in 2017. One would have thought Google would be leading the pack.
This revolutionary new technology poses an existential threat to Google Search. And a new development at OpenAI reveals what I’ve been predicting all along.
As a reminder, OpenAI is the company behind ChatGPT. This is the generative AI capable of producing content and even writing software code on command. It can also have intelligent conversations with us humans.
Well, OpenAI just released a version of GPT-4, the underlying large language model of ChatGPT, for web browsing. This is an absolutely radical change.
If we remember, ChatGPT was trained on data and events that occurred before September 2021. So any interactions with the AI were based on information prior to that date. The AI had no knowledge of anything that had happened since.
That just changed.
By enabling GPT-4 for web browsing, the AI can incorporate real-time data and information from the open internet. And that means it will be able to provide a far more robust search experience than anything we’ve ever experienced before.
To demonstrate this, let’s look at GPT-4 in action:
Here we can see a user instruct GPT-4 to list 10 things that happened in AI this week. Then the user asks GPT-4 to put these items into a table with links to the sources.
And sure enough, GPT-4 creates a clean table that’s easy to read in a matter of seconds. This is remarkable.
Compare this to the legacy search model. We would have to type our search into a web browser. Then we would have to sift through scores of links to determine if they were relevant to us.
It would likely take us at least half an hour to put together this same table – maybe longer. GPT-4 takes that process and makes it happen in a matter of seconds.
This is an incredible productivity tool. And it demonstrates clearly why generative AI is the next generation of search. There’s simply no comparison.
So I believe we will see Microsoft employ this technology through its Bing search engine in the weeks ahead. At that point, everyone will have access to it.
And that’s very bad news for Google. As a reminder, Google generates 80% of its revenue from advertising revenue. And that ad revenue is made possible by people using the Google search engine, collecting data on those users, and targeting ads based on that data.
But what do we think would happen to those advertising revenues if people abandon Google Search for a superior search product? We can see why this is an existential threat.
To make matters worse, Google is at least several months behind OpenAI and Microsoft on this technology. It may be as far behind as six months.
That may not sound like much… but it’s huge.
This technology is advancing much faster than anything we’ve ever seen before. In fact, OpenAI is releasing new versions of its generative AI every three months or so. And the rate of adoption by consumers of this technology is also the fastest rate of technology adoption in history. It won’t be long before a billion users are using generative AI for search.
This might be the biggest tech story of the year. Are we going to watch Google’s seemingly unstoppable business model crash and burn this year? Will we see Microsoft and OpenAI capture more than 1 billion users? Or will there be a new entrant with an even better generative AI that jumps ahead of them both?
An interesting development at Walmart just caught my eye. The retail giant teamed up with a company called Pactum AI to use artificial intelligence in its negotiations with suppliers.
This probably isn’t something we would expect to see, as the process of negotiations tends to be an iterative human process that takes time and technique. But that’s exactly the point. The human process is full of inefficient friction. AI can solve that. Here’s how it works…
Walmart gives the generative AI all the parameters around pricing, terms, and conditions that it deems acceptable with regard to its supplier agreements. Then the AI is empowered to negotiate with suppliers through a chatbot. The format is no different from communicating with ChatGPT.
And once the negotiations start, Walmart is completely hands-off. The AI has free reign to work and try to get a deal done. And the results are surprising…
As it turns out, the time to close a deal has dropped from weeks or months to just days. The AI-enabled process is much faster.
And get this – about 75% of Walmart suppliers say they prefer dealing with AI rather than a human negotiator. We can imagine that’s because the AI doesn’t leverage emotions or underhanded tactics. It’s pure business.
The AI also doesn’t have to “get back to us” with an answer. It doesn’t take vacations or get distracted with other tasks.
So AI is creating a stronger win-win relationship between Walmart and its suppliers. This is a bigger deal than we might imagine.
In my time as a corporate executive in Japan, I sold semiconductors to companies that were Walmart suppliers. Back then, those companies had to send representatives to Walmart’s headquarters in Arkansas to go through the negotiation process.
While they found the experience interesting with such a giant like Walmart, it was always a difficult and time-consuming trip. And often it was frustrating – partly because there was a language barrier between them and Walmart’s negotiators.
Generative AI changes that dynamic entirely. No more travel required. No more language barrier. No need to deal with human emotions or cultural differences.
Instead, suppliers can negotiate with Walmart’s AI right from a computer. What a remarkable shift.
This is a great example of one of my very favorite investment themes. Companies that develop technology that can remove friction from transactions of any kind have a strong tendency for fast adoption.
Pactum is still a private company having raised its series A2 round last December. It has strong potential and is backed by Jaan Tallinn, co-founder of Skype, as well as the venture capital arm of e-signature powerhouse DocuSign.
Speaking of AI for corporate applications, a team of researchers at MIT and Stanford just published the very first research paper about generative AI usage in the workplace. This is our first formal look at how use of the technology benefits a company.
The researchers set out to understand how generative AI is being adopted… and what kind of impact it is having at the company level.
To that end, the team analyzed the data of over 5,000 customer service agents. This information was used to train a generative AI in a way that was specific to an individual company. Then they put the AI to work. And what they found was that the introduction of AI increased productivity in customer service centers by about 14% on average. That’s measured by the number of issues resolved per hour.
They also found that the highest productivity increases came from novice workers. The more experienced workers saw smaller performance bumps.
This makes perfect sense. And it means that generative AI can essentially shorten the learning curve for newer employees. That provides companies with an immediate impact.
The other interesting thing this research team found was that customer requests that required management involvement declined once generative AI was involved. That means that customer satisfaction improved when the company’s customer service agents were empowered with the technology.
And get this – the use of the technology also increased employee retention. That suggests the employee experience improved as well.
This is big.
There tends to be high employee turnover in the world of customer service. It’s a stressful environment for many people… so it’s common for employees to get frustrated and resign.
So the big takeaway here is that deploying generative AI across customer service departments is a major win for all parties involved – companies, customers, and employees.
That’s why over 25% of large corporations that are currently using AI are using the technology in their customer service department. This stat comes from management consulting giant McKinsey.
And that means customer service is currently the number one corporate application for AI right now. It makes a lot of sense. Improving customer service is low-hanging fruit. Solving product or service issues for any company is largely formulaic with a limited set of parameters. And it is possible to determine a set of well-known best practices for problem resolution.
It will take time, but the shift will happen in stages. Customer service agents will be empowered with the technology, and performance will improve. Then the technology will be enabled for autonomous operation, effectively AI customer service agents, for a portion of customer service needs. And in time, all but a very small number of issues will be handled by an AI.
When the economic incentives are there and customer satisfaction improves, the barriers for adoption drop quickly, and it becomes a competitive necessity to make the jump to AI.
Regards,
Jeff Brown
Editor, The Bleeding Edge
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
