• How valuations are set in private markets
  • 1984 was supposed to be fiction
  • Visualizing the speed of improvement for ChatGPT

Dear Reader,

Welcome to our weekly mailbag edition of The Bleeding Edge. All week, you submitted your questions about the biggest trends in tech and biotech. Today, I’ll do my best to answer them.

If you have a question you’d like answered next week, be sure you submit it right here. I always enjoy hearing from you.

Who determines a private company’s valuation?

Jeff,

This is a follow-up on Lonnie’s question in the 4-7-23 mailbag. I agree with your strong emphasis on valuation. How is the valuation of private companies determined? 

Since there is little or no public information available, who is to say what the company is worth? Does the company determine the valuation, or do the VCs, or is it some other third party? Enjoy your Bleeding Edge. Thanks.

 – Iver L.

Hi, Iver. Thank you for the follow-up question on this topic. In hindsight, this would have been great information for me to share in response to Lonnie’s question, so I appreciate you presenting me with the opportunity to do so.

But before I dig in, I have to warn everyone that the answer might not be what you expect.

Some might answer the question in a simplistic way by saying that the valuation is determined by what price investors are willing to pay for it. And that explanation isn’t incorrect… But there is a lot more nuance to understand in the process of valuing a private company.

We’ll start with the easier category of well-established private companies with revenue, earnings, and free cash flow. Companies like these are relatively straightforward to value. A large investor would typically analyze the free cash flows of the business, discount those future cash flows based on forward looking assumptions, and determine a valuation based on those numbers.

A typical valuation process will also look at a competitive landscape to find comparable companies – or “comps” – that can be used as a reference for competitive valuations. A lot of time is spent on understanding why the valuation multiple for one company in the same space is higher or lower than others. Valuation multiples like enterprise value to sales (EV/Sales) or enterprise value to earnings before interest, taxes, depreciation, and amortization (EV/EBITDA) are often used and can vary significantly based on gross margins, revenue growth, and the overall growth outlook for a company.

Companies that are much earlier in their growth stages that may not yet have a product or product revenues are valued in different ways. It’s true that almost all early-stage companies still have revenue and free cash flow forecasts that can inform a valuation. But these are typically taken with a grain of salt. It is extremely rare that they are ever accurate. There are simply too many variables and assumptions that can impact the future outcomes.

Valuing early-stage private companies is very much an art form. Finding a valuation requires making many assumptions. It has to consider the amount being raised, the stage of the company (i.e. which funding round), the sector the company operates within, the expected growth of the company, future gross margins, and expected competitive positioning in the marketplace.

And there are also other larger, macro factors that come into play. When the economy is hot and private capital is investing heavily into risk assets, valuations expand like we saw in 2020 and into early 2021. And when the economy suffers and private capital moves to safety, we’ll see valuation multiples contract accordingly like we saw all last year and during the first quarter of this year. These valuation multiple expansions/contracts have nothing to do with the underlying company. But they do impact the valuations.

And how “hot” the sector is also greatly impacts valuation. For example, right now, generative AI companies are flying hot. Massive valuations with companies that have zero revenue. And when I say massive, I mean $1 billion plus valuations. It may seem illogical, and in most cases, it is. But if the company is successful and reaches a certain level of adoption with the business model that can be quickly monetized, sometimes those valuations are justified.

It’s also worth noting that if there is a lot of money chasing a single private company, the valuation will be driven higher quickly as private capital competes to get an allocation in a funding round.

At the end of the day, there tend to be two parties that need to come to agreement. The management team of the private company and the lead investor, typically a venture capital or private equity firm, need to come to agreement on valuation. And when they do come to agreement, they establish what’s known as a pre-money valuation. That’s the valuation at which investors in that round invest at. Post-money valuation is determined by taking the pre-money valuation and then adding the amount raised. For example, in a $25 million round, if the pre-money valuation is $100 million, the post money valuation will be $125 million.

Thanks again for the follow-up on this topic…

1984 was supposed to be fiction…

Hello, Jeff, love your newsletter, The Bleeding Edge. It’s the only way to keep up to date with so many things going on nowadays. 

Just finished the book 1984 by George Orwell, and wow. Throughout the entire book, many things remind me of what the government is doing now. This was a fictional story, right?

Now add in the “RESTRICT Act”, central digital dollar, social credit score, and carbon score (Master Card is already playing with this).

With all this in place, this book could become reality, and people could start getting "vaporized" if their score gets too far off. The book should be a must-read for anyone who wants to stay "human." 

 – Eric A.

Thanks for being a reader, Eric. I’m glad you’re enjoying The Bleeding Edge. And I emphatically agree with you! 1984 was meant to be fiction, and yet it is turning out to be reality before our very eyes.

In fact, there are great t-shirts out there right now that make this point very clearly.

The RESTRICT Act was a perfect example. It was pitched as a piece of legislation to ban TikTok. This sounds good in theory.

After all, TikTok surveils user activity on the smartphone or computer; including data that resides on the phone, GPS location, website history, and a wide range of data that most of us would not be comfortable sharing. And it all gets sent back to mainland China for analysis, storage, and utilization.

But the RESTRICT Act went well beyond simply banning TikTok. The legislation would allow the government the “right” to “desktop applications, mobile applications, gaming applications, payment applications, and web-based applications.” All it would take is for somebody to be loosely defined as a “threat.”

And as we’ve seen over the past several years, that could easily encompass professionals who are simply not conforming to the desired narrative.

It’s a similar story with a central bank digital currency (CBDC). There would of course be benefits. There would be less friction, and stimulus funds could be “airdropped” overnight.

But it would centralize even more power in the hands of the Federal Reserve and – ultimately – the federal government. A CBDC could create a situation where “undesirable” people are denied the ability to pay for certain goods. Or they could be “de-banked” with the click of a button.

It’s simply too much power. And it’s very scary.

There are glimmers of hope, though. There has been an incredible public uproar over the RESTRICT Act. And even a CBDC – which is less known by the public – is gaining political attention. The governor of Florida – for instance – is pushing for state legislation that would block the adoption of a CBDC in the Sunshine State.

And you’re right, Eric, 1984 should be required reading in every school. So should Orwell’s Animal Farm. That’s another one to read if you haven’t done so already. And if you really want to go down the rabbit hole, Ayn Rand’s Atlas Shrugged is an absolute masterpiece on human nature and that of societal development. It is deeply philosophical, and another great literary example of predicting what is happening today despite being written decades ago.

We are on the cusp of an epic struggle between those that want to impose oversight and tyranny upon populations, and those of us who want to live free and take responsibility for ourselves and our families. Many countries will succumb, others will divide, and some will collapse completely.

Freedom, and the right to live free, is a fight worth having. And I hope that there are enough of us that respect those rights and the rights of others to live free, that we can manage through this difficult period in history and leave the world in better shape for the next generations.

Did Microsoft “overpay” for ChatGPT?

Jeff,

I have been using the new Bing on Microsoft Edge and have found it really lacking. It is wrong more than 50% of the time.

I have asked it questions as mundane as net worth of celebrities, FDA positions, FDA guidance’s, chemical formulas, size of companies, and a long host of other query types. I then type the question into Google’s search in Chrome and usually get the answer. Many times, I know the answer and am just testing ChatGPT.

When I ask for a photo of an actor just to remember if I had it right in my mind, I get a response from ChatGPT that it is private information and it cannot comply. What? Anyway, Microsoft way overbought ChatGPT, and from my perspective Google has nothing to worry about.

 

– Ronald C.

Hi, Ronald. Thanks for writing in. I think that it is great that you’ve been experimenting with this incredible technology. As you’ve noted, it’s far from perfect, and there are clearly areas for improvement. But what’s exciting is that it is improving very quickly.

But before we dig in, there is one nuance to understand. These large language models (LLMs) like GPT-4 aren’t trained on up-to-the-minute information. In that way, they aren’t yet like a search engine. 

Here is a direct snapshot taken from OpenAI’s GPT-4 product announcement. Microsoft’s generative AI features are all based on this technology.

To catch readers up, Microsoft recently integrated the generative AI GPT-4 into its Bing search engine. In essence, ChatGPT is now “with us” when we use Bing.

It’s true that the AI isn’t perfect yet. I’ve written about this in the past.

Large language models like ChatGPT are trained on a massive body of knowledge, most of it from the internet. There is a lot of work for the industry to do here… after all, garbage in, garbage out. 

These are some of the limitations that the technology is facing right now. And the best LLMs will have a curated data set to improve the accuracy of the software over time.

But I don’t believe these nuances will stop – or even slow down – the adoption of artificial intelligence. OpenAI, the company behind ChatGPT, is adhering to an old Silicon Valley adage: Move fast and break things.

The company is moving very quickly right now. ChatGPT was only “unveiled” to the world last December. Four short months later and the technology is already being adopted widely. Here are a few items GPT-4 – the latest version of the AI – accomplished in just a two-week period.

  • It can write software code for just about any application.

  • It can create music based on any genre we like. We can just feed it the words we want.

  • It has created new extensions of internet browsers.

  • It can write and send e-mails for its users.

  • It can generate new business ideas.

  • It has been used to automatically configure and optimize software programs.

  • GPT-4 can now integrate with hundreds of software applications and use that software to perform tasks for the user.

  • It was used to hire a human worker to perform specific tasks.

  • GPT-4 can invent a new language.

  • It can now be used to hack computers.

  • It can design and code new smartphone applications.

  • It can design and create a website from a hand drawn image.

And the thing to keep in mind is not just how widely this technology is being adopted, but how quickly it is improving.

As a simple example, GPT-4 passed a bar exam with a score that was in the top 10% of human test takers. Of course, GPT-3.5 (ChatGPT) also passed the bar exam… but its score was in the bottom 10% of test takers.

So the AI went from the bottom to the top 10% in just over three months. That’s an incredible improvement.

And it doesn’t stop there.

OpenAI created a great chart that demonstrates the power of GPT-4 compared to GPT-3.5. Here it is:

Source: OpenAI

This chart compares GPT-4’s performance to that of GPT-3.5 across a wide range of academic tests. And we can clearly see that GPT-4 performed dramatically better on most tests.

In fact, GPT-4 more than doubled the performance of GPT-3.5 in many cases. That’s evidenced by the green bars being so much taller than the blue bars.

These are just a few examples of the rapid pace of improvement of the technology. It is getting better so quickly, it is hard to image what the performance will be like by the fall of this year. I can’t wait to find out.

Now let’s get back to your original question, Ronald, about whether Microsoft overpaid. There is some additional context that is important to understand.

Microsoft actually invested $1 billion back in the summer of 2019 when almost no one knew about OpenAI. I’m happy to say that Bleeding Edge subscribers were on top of these developments at the time.

I’m going to assume that Microsoft invested at roughly a $3 billion valuation in exchange for its $1 billion investment.

Then, most recently, Microsoft invested another $10 billion at a $29 billion valuation. That implies that its originally $1 billion valuation is up 9.66 times the original investment. That means that on paper, its original investment of $1 billion has now grown to almost $10 billion in the span of less than about three and a half years.

Assuming that OpenAI exits via a future IPO or acquisition at a valuation higher than $29 billion, Microsoft will make an absolute fortune. But let’s set that aside.

Let’s assume that OpenAI dramatically improves ChatGPT over the next year. GPT-5 will follow with GPT-6 after that, and each release will have noticeably better improvements. And new functionality will be added by companies like Microsoft to incorporate real-time web searches with the historical information that the LLM was trained on.

If 100 million people were to pay $20 a month for this incredible technology, that would result in $24 billion a year in annual revenue at very high gross margins. At a 5X enterprise value to annual sales, that would put OpenAI at a $120 billion valuation. But let’s set that aside as well.

By employing ChatGPT, Microsoft’s search engine has now reached more than 100 million users for the first time in history. What’s that worth? How much additional advertising revenue will Microsoft generate with all of these new users?

Microsoft made $11.5 billion in advertising revenues in 2022 fiscal year. How much additional advertising revenue will be made with all of the new users coming on board due to ChatGPT? Three billion dollars? Five billion dollars? More?

A strong argument can be made that on paper, Microsoft already made its money back from its most recent investment. And it doesn’t take much extrapolation to see that it will probably generate an additional $10 billion in advertising sales within the next two or three years.

In short, not only do I think that Microsoft didn’t overpay, but it made one of the best deals that the company has ever made in history. Not only will Microsoft not lose a dime, but it will also make a multi-billion-dollar fortune for this investment that will confer further competitive advantage to all of its other product categories.

As for Google, the company is in panic mode right now. It got caught flat footed. In the tech world, we don’t worry about how things are today (yes, ChatGPT isn’t perfect). We worry deeply about what things will be like in six months, 12 months, or two years from now. We have to anticipate future competitive dynamics and adjust our own strategies accordingly.

Google knows how quickly things are moving. If it doesn’t make rapid progress right now, Bing’s 100 million users could quickly turn into 500 million users and erode Google’s market share.

Either way, this will be an exciting battle to watch unfold in The Bleeding Edge. I hope my perspective is useful in stepping back and looking at the bigger picture of a very interesting – and very large – investment.

Regards,

Jeff Brown
Editor, The Bleeding Edge