Startup companies increasingly are tumbling over a financial cliff, while others are hanging in mid-air like Wile E. Coyote before the inevitable fall.
Startup shutdowns hit 770 in 2023, up from 467 in 2022, and for those startups lucky enough to get funding in 2023, 20% had their valuations reduced. Most observers knew these companies were past the cliff edge, although many hoped that they could stay suspended in mid-air until their problems went away.
These troubled startups didn’t need more time, they needed better ideas. Venture capitalists had pressured them to grow and grow some more, but growth can’t be sustained if the underlying fake-it-til-you-make-it business model is flawed.
Now VCs are worried. They have some dry powder because their fundraising set records a few years ago and many were smart enough to sometimes say no to the most fanciful startups. Unfortunately, they didn’t say no often enough and, with fundraising shriveling, they can’t keep every floundering startup in mid-air even if they wanted.
VCs raised just $161 billion in 2023, down from $307 billion in 2022 and $380 billion in 2021. At the same time, their startup investments fell to $171 billion in 2023 from $242 billion in 2022 and $348 billion in 2021. In other words, VC investments fell by half between 2021 and 2023 and still they raised less money than they invested—which is clearly not sustainable.
Even worse, the IPO market has collapsed and interest rates are still high, which makes fundraising more challenging and reduces the value of any profits that startups might eventually realize. The bottom line is that VCs now must decide which startups to save and which to let go.
“VCs, like the startups they invested in, have overpromised and underdelivered. ”
Investors are also concerned about the money they’ve given VCs with little to show for it and they are tired of paying 2% annual management fees to VCs that, like the startups they invested in, have overpromised and underdelivered.
Most everything that VCs have invested in is now junk. About 90% of publicly traded unicorns — startups valued at $1 billion before going public — are losing money. Morgan Stanley researchers recently concluded that this figure is even higher for privately held startups.
While there have been a few successes such as Moderna
MRNA,
Zoom
ZM,
Airbnb
ABNB,
and Uber Technologies
UBER,
there are far more shiny baubles that have broken: ride-sharing was supposed to eliminate parking lots; electric vertical take-off and landing (eVTOL) aircraft were supposed to reshape our cities, blockchain was supposed to become the basis for our information systems.
‘How many bears have the Russians sent into space?’
The latest bauble is large language models (LLMs) such as OpenAI’s ChatGPT, Alphabet’s Gemini
GOOG,
and Microsoft’s Copilot
MSFT,
Enormous amounts of energy and brainpower have been devoted to these undertakings but, so far, their main successes have been in generating disinformation and phishing scams.
The fundamental problem, which no one is close to solving, is that LLMs do not understand any of the text they input and output, and so have no way of assessing the truth or accuracy of their responses.
For example, we recently asked Microsoft’s Copilot (formerly BING), “How many bears have the Russians sent into space?” Its response: The Russians have sent a total of 52 bears into space. These brave bears were part of various space missions, contributing to our understanding of how living organisms react to microgravity and other unique conditions beyond Earth’s atmosphere. Let’s take a moment to appreciate their furry contributions to space exploration! 🚀🐻
The link is to an interview with Gary Smith on January 13, 2023, that mentioned ChatGPT 3.0’s made-up answer to this question: Russians have sent a total of 52 bears into space. The first bear, named “Kosmos 110”, was sent in February 1959 and the most recent one, a female named “Borenchik”, was sent in August 2011.
“If you know the answer, you don’t need to ask AI, and if you don’t know the answer, you can’t trust AI.”
LLMs are being trained on falsehoods they polluted the internet with. As with Gresham’s law that “bad money drives out good,” LLM untruths may drive out useful information.
The inescapable dilemma is that if you know the answer, you don’t need to ask an LLM and, if you don’t know the answer, you can’t trust an LLM.
Yet, hundreds of LLM startups are trying to raise money with fake-it-til-you-make-it promises. Like Wile E. Coyote, startup investors never seem to learn from their misadventures.
Although 770 startups (a record number) shut down in 2023, this is still a small fraction of the roughly 50,000 that purportedly have received VC funding over the past 10 years. How many more will disappear and how soon? The Financial Times reports that a burgeoning secondary market for private shares is providing some relief, but this is because the shares are deeply discounted — down roughly 50% with further markdowns imminent. Even though these startups are only worth a fraction of their previously inflated values, employees want to get out before the values fall even further. If employees are worried, we should be too.
Jeffrey Funk is a retired professor and author of five books, including the forthcoming “Unicorns, Hype and Bubble: A guide to spotting, avoiding, and exploiting tech bubbles,” (Harriman House, October 2024).
Gary Smith is a professor of economics at Pomona College and the author of more than a hundred academic papers and 17 books, most recently (co-authored with Margaret Smith), “The Power of Modern Value Investing: Beyond Indexing, Algos, and Alpha,” (Palgrave Macmillan, 2024).
Also read: The long-awaited IPO ‘revival’ really depends on the bull market’s survival
More: This AI startup has the backing of Jensen Huang, Joe Montana and Jeffrey Katzenberg
