Description

We are in the midst of the final innings of one of the largest financial bubbles in recent history, artificial intelligence.

Nvidia sits at the center of this bubble, seemingly poised to benefit from the boom in AI research and development.

Unfortunately, history suggests that this time is likely not different. Artificial intelligence tends to progress in waves of ebullience and doldrums. An advance is made and people extrapolate it into the future, assuming that Artificial General Intelligence is only several years away.

“In 1970 Marvin Minsky told Life Magazine, “from three to eight years we will have a machine with the general intelligence of an average human being.” However, while the basic proof of principle was there, there was still a long way to go before the end goals of natural language processing, abstract thinking, and self-recognition could be achieved.”

https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/

While some of the recent advancements in AI are no doubt exciting and represent great progress, we believe the market is currently wildly too optimistic about the future pace of progress in AI and that this cycle will be no different than past AI hype cycles.

ChatGPT is an exciting product, but internet searches suggest that its growth has stalled out. While OpenAI does not release any usage stats, we believe it’s likely that the retention data of ChatGPT’s users is bad, and growth/recurring usage are down considerably.

As a proxy for ChatGPT usage, we pulled the search trends for the term ChatGPT Login, which we believe represents people trying to login to the service - it shows that usage is well off the highs, suggesting some of the initial enthusiasm has worn off. While it’s a neat tool, as most have seen, its accuracy causes great concern given the high error rate, and the true value may be a lot less than many initially expected.

How does this all relate to Nvidia? While NVDA is up ~200% YTD, NVDA revenues actually declined 13% YoY in its most recent quarterly results. But why then is the stock performing so well? Expectations for massive future growth, driven largely AI orders of Nvidia chips specially designed for AI use cases.

Nvidia projected $11bn of revenues for the 2nd quarter of the FY 2024. This is in large part driven by a wave of capex from AI companies. We believe it’s likely much of these orders are one-time in nature given the companies putting them in are burning cash at astronomical rates. It seems the market has forgotten the lessons of the past few years. OpenAI reportedly lost $540mm in 2022 

https://www.theinformation.com/articles/openais-losses-doubled-to-540-million-as-it-developed-chatgpt

To help fund this capex by a range of nascent companies, Nvidia is investing its own capital in many of the same companies that are putting in orders of the chips. 

6/8/2023:

June 8 (Reuters) - Cohere, an AI foundation model company that competes with Microsoft-backed OpenAI, said on Thursday it had raised $270 million in a funding round from investors including Nvidia (NVDA.O), Oracle (ORCL.N) and Salesforce (CRM.N) Ventures.

6/13/2023:

Artificial intelligence-based video generation platform Synthesia has raised $90 million from investors, the company told CNBC exclusively.

The round, which values the company at $1 billion, was led by venture capital firm Accel and backed by U.S. chipmaker Nvidia.

How does the funding mechanism work? Nvidia partners with leading VC firms to invest in AI startups which then take that capital and place orders with NVDA for chips. This helps boost NVDA’s stock price given NVDA currently trades at a Price to Sales Multiple of ~25x. So every $1 of Sales generated by startups buying NVDA’s chips generates $25 of value for NVDA stock.

Like most bubbles, we believe this is unsustainable. Eventually the AI startups need to generate cash flow or they cannot continue to massively invest in Nvidia chips. Past chip cycles like Crypto Mining have tended to peter out over time as they were fundamentally unsustainable. We believe the same will hold true this time.

Nvidia currently trades at a valuation not even matched by the excesses of the 2000 Internet Bubble. Even if AI ends up being as transformative as experts are predicting, it will take far longer than expected and involve numerous bumps in the road. There is also the possibility that there are technological advancements that displace Nvidia as the leading Chip solution. The first mover in a space is rarely the last mover as we have seen numerous times in the past. Check Excite and Yahoo! For some cautionary tales on the risks inherent in being first out of the gate in a hot new area with new entrants chomping at the bit.

As a reminder, this is far from the first time Nvidia has seen a short-term boost due to a temporary dynamic. Nvidia saw a revenue boost from crypto mining because its chips were the best suited to crypto mining. This ended up unwinding: Nvidia’s crypto-associated revenues went down ~80% from 3Q21 to 4Q21:

https://decrypt.co/93106/nvidia-crypto-mining-processor-revenue-slumps-77-24m-q4-2021

We believe the same dynamic likely will occur with AI.

Another interesting point which has gone largely unnoticed is Nvidia insiders have been selling large blocks of stock in recent weeks. Just since May 31, insiders have sold >$100mm of stock. We believe this is a very telling sign - these are not people who need to sell stock to buy a home or renovate their house - these are insiders selling substantial portions of their holdings likely because they recognize there is an AI bubble going on.

Catalyst

father time is undefeated

eventual capital raise