It has never been cheaper to get AI. AI funding is up, and so is revenue, but what if costs keeps increasing, will the bubble pop?
What is a Bubble?
Historically, bubbles have been characterized by exponential growth in investment with lagging revenue growth. In plain English: a lot of money came in, and almost nothing came out. This happened during the dot-com bubble of the early 2000s. We expected the internet to fuel commerce. Eventually, the internet did give us Amazon and Alibaba, but the first winners were Google and Facebook. So search, and social, came before commerce. Yet, the only winner during the dot-com implosion was a hardware company called Cisco. Cisco stock price soared to 3,800% from 1995 to its peak in March 2000, making it the most valuable company globally at that time. The same happened to NVidia, another hardware company. What are the odds of history repeating itself?
Dot-com was not the only bubble. We also had Covid-19, which brought us low interest rates and shelter-in-place behaviors. This drove investments in companies sky-high. Once life returned to normal, all valuations came crashing back to Earth. No other company exemplifies this more than DocuSign. Due to its rapid rise during the pandemic, Docusign was included in the Nasdaq-100 in June 2020. However, its listing lasted two years, and it got kicked out by December 2022. Yet, Docusign is one of our most reliable canaries.
So the question is, are we in an AI bubble? Let’s calculate some numbers and look at the graphs.
What is Venture Capital telling us?
October 2024, we registered a record five fundraise announcements in a single day. This year, we witnessed a billion raised by legal startups in a single day, another record. We made a video chronicling all historic legal fundraises over the years starting in 1984 with Adobe. The company that created PDF: the only way legal could replace paper to securely share contracts. Bringing legal from stone tablets into the digital age. Sorry, we digress. History is the past, and we’re now in the present. So how does funding look in 2024?
Glad you asked, since we have a few more records to announce. The graph above shows us the total raised by legal companies in the past six years in each quarter. You’ll spot that (a) Q3 2024 is the highest ever Q3 total and (b) ranks third overall after Q1 2021 and Q1 2022. Yet, as we reported in reaction to investment bankers, the total number of companies getting funding is dropping. Translation: more money went to fewer companies. This brings us to another new record: at no time in recorded history have legal companies raised this much money on average. Investors clearly feel confident that they are picking winners. Same winners as in the Dot-com and Covid-19?
We mentioned bankers, which brings us to another metric most overlook: Debt. This year, we are on track to set a new record in bankers offering credit to legal ventures. Same as funding, fewer companies received more in loans than at any time in history. That is $4.1 billion in credit which is up 130% over last year. Even with interest rates on loans at an all-time high, bankers believe in legacy companies to survive. If you are interested in our take, read our rundown of them all here: Data, Distribution & Deals. Spoiler: we disagree. We could be wrong, and here’s why.
What is AI revenue telling us?
The first line in this piece ends with:”…lagging revenue growth“. The revenue generated from AI applications in the legal field is the critical metric. A simple proxy to calculate growth is OpenAI revenue. It has been the fastest company in history to a billion in annual revenue. That’s why it is now arguably the highest-valued private company in history. However, it is also the most capital-intensive company ever and projected to be profitable by 2029. Meanwhile, their expenses will triple to $14 billion by 2026. But who is receiving $14 billion from OpenAI?
In hindsight, history had already answered this question, yet it took us two previous posts to unravel. The answer: producers of general processing units (GPU) and electricity. Now, a scientific breakthrough in either will trigger the biggest wealth destruction we will ever witness. That is the Black Swan scenario. Meanwhile, all AI companies, including legal, are funneling money towards GPUs and energy. Economists at Goldman Sachs calculated about 4–16 months of runway to keep spending on AI. If earnings do not keep up, how can one keep paying expenses?
Here’s a thought experiment: there is a nuclear power plant that does not produce power because it is closed due to the biggest nuclear disaster in the Americas. Here’s a pop quiz:
- Why did a software company (Microsoft) buy this nuclear power plant (Three-mile Island)?
- Maybe because Microsoft sees energy as its new source of revenue?
- Perhaps software revenue will decline over time?
- Is it that AI, will eventually write all software?
- Is Software-as-a-Service (SaaS) dead?
What revenue is telling us, is to look at the costs. With AI, the production costs are concentrated. If we could only decentralize and distribute the AI costs. Where did we hear this before? Oh, I remember: Why the World Needs Decentralized Legal AI.
Author note: this may be my last post. AI will flood the airwaves, so posting will become futile. I wrote these posts as a diary, to remember.