OpenAI funding round values company at $157 billion

[Daniel Nenni]

Leadership turmoil, rumblings about an AI bubble, and a failed CEO coup last year didn't stop OpenAI from raising $6.6 billion.

On Wednesday, OpenAI completed its latest funding round according to the New York Times, giving it a valuation of $157 billion. That makes it the largest VC round of all time, a distinction that was previously held by Elon Musk's xAI, which raised $6 billion earlier this year.

The investment was led by VC firm Thrive Capital and included Microsoft, NVIDIA, Japanese tech investment company SoftBank, and United Arab Emirates investment company MGX. As confirmed in previous reports, Apple, which was rumored have backed out of investment talks with OpenAI is not an investor.

OpenAI's massive funding round caps off a raucous period of events over the past year or so. After it ignited the generative AI boom in 2022 with the launch of ChatGPT, OpenAI quickly became the latest Silicon Valley darling, earning $13 billion in investment from Microsoft. Then last November, a failed attempt by its non-profit board to oust CEO Sam Altman, revealed leadership's mistrust of Altman and differing philosophies from its co-founders. Eventually, co-founder Ilya Sutskever and, just last week, CTO Mira Murati quit the company. Co-founder and President Greg Brockman has taken a leave of absence and other executives have recently quit the company. OpenAI is also in the midst of restructuring its corporate status from non-profit to for-profit.

Despite waning investor interest, backlash from AI-weary consumers, and unfulfilled promises of AI's potential suggesting that the AI bubble might be ready to burst (or perhaps that it's slowly leaking after bursting months ago), OpenAI's funding round has proven investors with deep pockets are willing to bet on its success.

OpenAI makes money from subscriptions to its ChatGPT Plus, ChatGPT Enterprise, and API accounts. As of August, OpenAI reported that ChatGPT has over 200 million weekly users, which was double the amount of users in November, 2023, and expects to bring in $11.6 billion in sales next year (per CNBC). But building and maintaining AI models isn't cheap. According to The Information, OpenAI could lose up to $5 billion this year.

MSN

 

[Daniel Nenni]

Is AI a bubble?

Dot-com bubble - Wikipedia

en.wikipedia.org

 

[tooLongInEDA]

Is AI a bubble?

Dot-com bubble - Wikipedia

en.wikipedia.org

Yes, if we follow that article. And apply a little common sense to what's happening today (collosal investments with only marginal returns so far; massive increase in energy consumption from AI data centres - in part due to non-market pricing of services today; too many players crowding an emerging market, yet all priced to win; domination of US market by a handful of tech stocks; extreme stock volatility).

And things like yesterday's post about Cerebras' planned IPO. I don't know their technology and whether it's any good. What struck me was that they're still losing money and 83% of their sales are to a privately held company based in the UAE. There used to be slightly tougher criteria for IPOs ...

I suspect that part of the cause here is the artificially accelerated growth that tech companies get in the "free stuff" era where they are able to under-price their services (and burn billions of investors cash) in the rush to be first/biggest. I'm fairly sure that - to take an example - the railway boom investors only got to subsidise the track and rolling stock and not also the ticket prices. It's almost like we've set out to build an unstable positive feedback system.

But a bubble of some sort may all be "normal and expected". Read the article's final words (under "Legacy"):

In a 2015 book, venture capitalist Fred Wilson, who funded many dot-com companies and lost 90% of his net worth when the bubble burst, said about the dot-com bubble:​

​

A friend of mine has a great line. He says "Nothing important has ever been built without irrational exuberance." Meaning that you need some of this mania to cause investors to open up their pocketbooks and finance the building of the railroads or the automobile or aerospace industry or whatever. And in this case, much of the capital invested was lost, but also much of it was invested in a very high throughput backbone for the Internet, and lots of software that works, and databases and server structure. All that stuff has allowed what we have today, which has changed all our lives... that's what all this speculative mania built.​

Aside: I think we all know there must have been a transistor boom/bubble in the 1950s. Was there really an accompanying stock market bubble then ?

Historically, the dot-com boom can be seen as similar to a number of other technology-inspired booms of the past, including railroads in the 1840s, automobiles in the early 20th century, radio in the 1920s, television in the 1940s, transistor electronics in the 1950s, computer time-sharing in the 1960s, and home computers and biotechnology in the 1980s​

 

[Daniel Nenni]

There have also been automotive bubbles where we have had dozens of car companies only to consolidate to a half dozen. We are going through that again now with EV companies. I was told at a conference last year that there were more than 300 automotive related start-ups in process (a lot in China). There is no way the automotive industry can support that large of an investment.

 

[tooLongInEDA]

Thinking a bit more about the nature of over-investment in AI (I'm assuming there is - just my opinion). This doesn't matter quite so much if what you end up with is an asset of lasting value, as was the case with all the dark fibre after the 2000 tech bubble. In the AI case however I suspect it's different. Chipsets which don't gain market traction may quickly become worthless once the software and support ecosystem goes. And those which do depreciate as quickly as any other ICs as newer, better versions are designed and built. Over-investing in telecoms capacity didn't matter so much as the kit is built to international standards and is largely interoperable. And it depreciates far more slowly.

In summary, the potential for massive write-offs and losses seems much higher this time around.

 

[adramtech]

Not a bubble, just a normal overbuild and big digestion period.

 

[soAsian]

Is AI a bubble?

Dot-com bubble - Wikipedia

en.wikipedia.org

The dot-com bubble may have burst, but it gave rise to tech giants like Amazon, Google...etc.

at least we know, Nvidia's capability for parallel processing in their chips holds potential for other applications (maybe?)

Intel kicking themselves for not getting lucky like Nvidia.

 

[Brady]

Not a bubble, just a normal overbuild and big digestion period.

I think Dylan Patel of SemiAnalysis on a podcast said similar, that in this AI investment cycle there’s been something like $60B in investment while the Dot Com cycle was something like $125B.

Definitely going to be huge amounts offline over investment, and so whether it’s a bubble or not is just a matter of semantics.

 

[Daniel Nenni]

I think Dylan Patel of SemiAnalysis on a podcast said similar, that in this AI investment cycle there’s been something like $60B in investment while the Dot Com cycle was something like $125B. Definitely going to be huge amounts offline over investment, and so whether it’s a bubble or not is just a matter of semantics.

The AI investment is not done. Reading about the Dot.com bubble and experiencing it first hand is two very different things. A lot of companies pivoted to Dot.com just like they are doing with AI. We all saw it coming, it was like a train wreck in slow motion. The actual losses are incalculable. AI is deja vu all over again. Just my opinion of course.

 

[Brady]

The AI investment is not done. Reading about the Dot.com bubble and experiencing it first hand is two very different things. A lot of companies pivoted to Dot.com just like they are doing with AI. We all saw it coming, it was like a train wreck in slow motion. The actual losses are incalculable. AI is deja vu all over again. Just my opinion of course.

I can imagine. I’m living through the AI boom first hand on the software side, and while I don’t have a comparison point to Dot Com, there’s real value being created by these model innovations this time. Of course there’s a ton of me-too’ing happening everywhere as well… ya know, there’s some Googles and there’s a lot of Pets.com.

Models of all kinds have not peaked in quality yet and there’s line of site to step-change improvements for at least year or two more. That will keep feeding those training runs, and keep fueling those data center buys, and so will keep fueling semiconductors.

 

[Daniel Nenni]

I can imagine. I’m living through the AI boom first hand on the software side, and while I don’t have a comparison point to Dot Com, there’s real value being created by these model innovations this time. Of course there’s a ton of me-too’ing happening everywhere as well… ya know, there’s some Googles and there’s a lot of Pets.com.

Models of all kinds have not peaked in quality yet and there’s line of site to step-change improvements for at least year or two more. That will keep feeding those training runs, and keep fueling those data center buys, and so will keep fueling semiconductors.

AI will definitely have a long reach but it is early yet. Even Sam Altman said AI software is bloated and needs to be optimized. Hopefully we will not need thousands of data centers because if we do we will need alternate sources of energy to power/cool them.