AI Laundromat
Audio Recording by George Hahn
AI Laundromat
I’m at Cannes, hungover, as I went to Yahoo Beach and saw the Chainsmokers. Cannes is everything I wanted in my twenties and thirties but didn’t have access to, since I was working all the fucking time and had no money or influence. Better late, I guess. Anyway, the two biggest stories of this year’s Cannes advertising conference will be:
1. The Musk/Yaccarino apology tour, which received a slightly cooler reception than if Milli Vanilli showed up at Spotify Beach.
“No Elon, really, you can go fuck YOURself.” –The Ad Community
2. My friend Michael Kassan hosting the kickoff dinner for 3C, his new firm, on the terrace overlooking the party for MediaLink, his former firm. Pro tip: CEOs with ovaries don’t do this shit. The definition of a “dick” move. I said this (verbatim) on a panel with Michael yesterday and he responded “You mean, a big dick move.”
But. I. Digress.
Vision Thing
The best founders articulate a vision of the future and put their company’s business model at the heart of it. It’s no different from Hollywood — the sets and costumes change, but the promise of a utopian/dystopian future is a constant. BTW, the headline that best depicts our world would be “On Average, Things Modestly Better Today, Globally.” But that’s click repellent. Ironically, the hot movie trend seems to be biopics about business, specifically successful products. Studio execs recently greenlit movies about Blackberry, Tetris, the Air Jordan, and, presumably after edibles, spicy Cheetos and Pop-Tarts. As the financialization of everything, which has CEOs signing breasts, becomes a Category 7 hurricane, expect an onslaught of movies about AI as übervillain.
In the latest Mission: Impossible installment (Dead Reckoning Part One), the villain is a sentient AI entity called … wait for it … “the Entity.” I’m working on a film myself, and, given my Hollywood career, it will be in theaters soon. Maybe.
AI Rewrite
In America, money is relevance — and the most relevant thing in the world is now a firm that designs GPUs. Nvidia has registered the greatest market cap growth in history ($2 trillion in the past year alone), becoming briefly the most valuable company in the world, as its chips are the literal heart of the AI revolution. No storytelling required. I appreciate that the folks from NVDA don’t come to Cannes, unlike GOOG/META, to run their fingers through the hair of media execs before shooting them in the face. My favorite Cannes moment: Sheryl’s 2013 book-signing, which attracted hundreds of female execs even as her firm was depressing millions of teen girls.
Back to NVDA. Think about this: One company has added the value of the global auto industry and the GDP of Sweden in 12 months. OpenAI, and by extension its sugar daddy, Microsoft, are similarly benefitting. But investor hunger for AI stories can’t be sated by a few businesses. CNBC makes an annual list of the 50 most disruptive companies, and two-thirds of the entrants on the 2024 list “describe artificial intelligence as ‘critical’ to their businesses.” The key word in that sentence is “describe.” Every comms exec and CEO who can spell “AI” (i.e., all of them) has decided Ai is the protagonist of their company’s story. However, similar to Game of Thrones, a lot of these leading men and women aren’t going to make it to Season 2.
Which brings me to Tempus. I mean … Tempus AI.
Tempus is a genomic testing and data company. Doctors take blood or tissue samples from patients and send them to a Tempus lab, and Tempus sends back information about the genes it finds in those samples. In addition to testing, Tempus licenses the data it collects to pharmaceutical companies, who use it to develop drugs. Tempus specializes in cancer, which is a great business, as fighting cancer is a noble thing. AI, healthcare, cancer … disco. I emailed the CEO — whom I (sort of) know; strikes me as an impressive guy — and asked if I could invest. He said they are only letting institutions invest (makes sense). I was still interested, however, so I looked further into the firm (read: I asked my team to look into the firm), and here’s what we found:
It’s not such a great business in the sense of making money. Tempus doesn’t make money, it burns it. Since its founding in 2015, the company has raised $1.5 billion in venture money and spent almost all of it. According to its IPO filing, Tempus had $80 million left in the bank on March 31. And it was burning $8 million per week, suggesting it would run out of cash … last week. The company raised an additional $200 million from Softbank in late April, otherwise it might not have made it to the IPO. I’m wondering when the CIA is going to plant Softbank, Chamath Palihapitiya, and Cathie Wood in Moscow to take the Russian economy down.
Tempus AI stumbled across the IPO finish line and fell into a pile of money. It priced at $37, opened at $40, and hit $44 before settling at $38, giving the company a $6 billion market cap after a day of trading. A respectable IPO bump, and a check for $411 million (aka a year of burn). The valuation was down from Tempus’s private market valuation of $8 billion, but an 11x revenue multiple is still greater than that afforded its competitors, such as Guardant and NeoGenomics, which trade at 6x and 2.7x, respectively. (Guardant welcomed Tempus to the public markets with a patent lawsuit three days before the IPO. See above: dick move.)
How did a money-losing business facing a patent lawsuit in a competitive market, run by a guy whose previous company, Groupon, trades at 6% of its IPO valuation after burning through $1.5 billion of investor capital, garner a multiple nearly double that of its most richly valued competitor? A: Never underestimate the market’s ability to provide a product/story when people have cash in hand. In this instance it will likely (again) be investors who get beaned in the face.
Tempus refers to AI 228 times in its IPO paperwork, even tacking those letters onto the end of its name last year. Notably, the company first filed for an IPO in 2021, and back then it mentioned AI 78 times. The AI hype refers to the third leg of Tempus’s stool, its “AI Applications” product line. The idea is that Tempus will combine its lab testing with a comprehensive review of a patient’s entire record — other test results, physician’s notes, family history, medications, etc. — and provide recommendations for patient care. An AI doctor. Which sounds amazing, but is also sci-fi (i.e., fantasy). Which it is, because the AI Applications segment currently provides 2% of the company’s revenue. Tempus hasn’t had time to add “AI” to its logo, and the only part of its website that incorporates the new name is the Investor Relations section. The S-1 reads like a venture capital pitch deck written by an LLM with the following prompt: “Pull together a 30-minute slide presentation for an IPO roadshow that positions us as an AI firm.” Second Prompt: “More cowbell,” if cowbell is AI.
Spin Cycle
I can’t decide if I should criticize Tempus or commend it. The market wants AI companies, Tempus wants the market’s capital, and it pairs the trade via AI washing. Everybody’s doing it, and the company’s ability to attract cheap capital may provide the steroids to turn it from Carl Lewis to Ben Johnson. Tempus AI isn’t the first company to play the name game: C3.ai started life as regular “C3,” had a cup of coffee as “C3 Energy,” and jumped on the “internet-of-things” bandwagon as “C3IoT” before going public as C3.ai.
In the U.K., the largest domestic energy company, Octopus, has jacked its valuation nearly 2x since 2001, and its CEO can’t stop talking about AI. Starbucks is using AI to “nurture the human spirit,” Kellog awards an “MBAi” business school degree. At my online ed startup, Section, we offer an AI Academy, but we haven’t changed the name to Section.ai. (Although GoDaddy is selling the URL for just $798,888 … a small price for a 2x valuation bump.) Even actual washing machines are in on the game: LG introduced an AI-powered washing machine in 2020. Goldman bankers bring a washing machine to every road show.
We’ve been here before, and the cycle always turns. Most recently it was crypto, and the NFTification of everything. The dot-com boom in the late ’90s saw the launch of businesses including DrKoop.com, the website of former Surgeon General C. Everett Koop, which popped 38% on its Nasdaq IPO before going under in 2001.
There are some signs the golden age of AI washing is slowing from the spin cycle — coming to an end. Regulators are paying attention. The SEC hit two investment advisers with six-figure fines for falsely claiming to use AI in their financial forecasts. And the agency’s enforcement head made it clear this was a warning shot for publicly traded companies: “Public issuers making claims about their AI adoption must also remain vigilant about similar misstatements that may be material to individuals’ investing decisions.”
The SEC also brought fraud charges against defunct recruiter Joonko, which claimed to use AI to identify diverse applicants, but flamed out last year when its founder was revealed to have inflated its numbers and concocted fake testimonials. The FTC wants companies to know that it’s watching their AI claims, and the FDA is looking into regulating the use of AI models in healthcare. The market’s favorite citizen-sheriff, Hindenburg Research, recently pointed its short-selling guns at Equinix, accusing the data center provider of selling an “AI pipe dream.”
Tempus AI’s IPO may be the latest signal that the AI washing cycle is ending, as its modest, pre-orchestrated pop has evaporated. And C3.ai’s stock is already off 80% from its post-IPO high in 2021.
Air Dry
The line between AI opportunity and AI washing is neither clear nor fixed. Sure, it’s obvious for the outliers — Nvidia will continue to register upside powering AI, and shady brokers who claim to use AI to pick stocks will not. But for most firms, clarity will only come with hindsight. Ironically, a year ago the big story was how to detect if someone was using AI (news stories, student papers, lawyers). Today, we’re attempting to discern if a firm is not using AI.
Life is so rich,
P.S. This week on Prof G Markets, Ed and I spoke with Ray Dalio, founder of the largest hedge fund in the world, Bridgewater Associates. Listen and subscribe here.
P.P.S. Want another take on The Science of Success? Section’s CEO is sitting down with Amherst professor Catherine Sanderson to discuss it on July 10. RSVP here.
Dear Scott
U R spot on about a global craze about “marketing and milking AI “by all sorts of ” intelligents ”
Obviously it will ultimately cost to gullible investors aka fools who either themselves or through their mutual funds will burn their hard earned savings which will get accumulated by the super intelligents in Global Economic System of present world .thanks for your attempt to create global awareness for innocent investors .
Of course I am sure some ” intelligents” will also find money making opportunity in this article ( like Public law suit lawyers /short selling experts etc ) but none the less this is quite interesting attempt by U .
Keep it up dear good human.
Prof .CA Nandkishor Pophale, Mumbai
I sort of doubt this data exists, but it would be really fascinating to see mentions of “Internet” from the late 90s overlaid on the AI mentions chart!
Thank you for your keen insight.
Testing
Hi Scott… thanks for the content.
I was hoping that ketamine helped you overcome your Musk Derangement Syndrome. Did you watch the super heavy launch?
It’s a bit ironic that you seem to suffer from depression but also penned “The Algebra of Happiness”. Your postings are riddled with drug use references (SSRIs, cannabis, TRT, etc…), and yet you preach that you have the formula for happiness.
The biggest difference between now and the late 90’s is that back then, tech was 40% of the S&P but it was a bunch of companies many of which lost money. Now it’s just one (essentially) and they DO make money. So this could last a while (who the F knows but keeping up with the indices is a bid draw….
They need to rename AI and call it ALDS – artificial librarian and data sifter. That is all it is. All the AI bullshit is designed to impress the rubes. There is no I in AI.
THIS is the best line – of the week at least – “I’m wondering when the CIA is going to plant Softbank, Chamath Palihapitiya, and Cathie Wood in Moscow to take the Russian economy down.”
I was thinking the same thing –to use our Hollywood term — foreshadowing: Softbank is the foreshadowing king of imminent fiscal disaster.
Still amazed that VC’s and such can be wrong 80% of the time and still have a job. And collect 2% fees.
There needs to be a “Mendoza Line”, seriously.
Now back to our regularly scheduled program…
S – check out Colin Chapman who founded Lotus . Formula One seems to have forgotten him . Crying out for a re-look . Apologies if you are cognisant of him . Plus you should interview Sir Jim Ratcliffe of Man Utd . Hession
Brilliant. I wish that more of my cohort would wake up.
Terrific article again….hope you run into my tenured professor at NYU one day, He is certainly not in your field or area of expertise but great in his own.
One of your best lines ever: “I’m wondering when the CIA is going to plant Softbank, Chamath Palihapitiya, and Cathie Wood in Moscow to take the Russian economy down.” Chamath should be in jail for pushing SPAC’s in 20 and 21. I can’t even imagine how much money people (like me) lost investing in the shit he pitched. He was on CNBC nonstop during that time and he (shockingly) has disappeared from the screen.
Pop-Tarts was done. “Unfrosted” with Jerry Seinfeld and Hugh Grant. I gave it a 1 star!
Once the dust settles, there will be plenty of legitimate AI use cases come along that change the game by increasing the speed, efficiency, and efficacy with which old problems are solved. Most of our problems are age old problems and the incremental gains we make in solving them are a function of ROI – how many resources can you deploy and get a 3 to 5 year payout. If we can train our models on these problems and deliver 100X the benefit with 1/100th the resources, we’ll see huge benefits from this technology. As long as it doesn’t fall into Gabriel’s hands, we’ll be fine. Wait – there aren’t any bad actors, right?
I believe early investors in Groupon later referred to it as Strapon.
Hey Scott. You should have taken this week off and joined 4% of the Scottish population in Germany having the time of our lives and making memories that will last a lifetime with friends and family. The soft-power effect of the Tartan Army is insane.
Love your takes. Would like to hear you talk about how to retire with money in the market. 1.5 million. Do you use cheap exchange funds. I have a broker from Merrill Lynch. What advice going forward. 65 years old and retiring in a year or two
Where should my money be. Apple Nividia and google are my biggest stocks. Thanks
Welp, that was fast. Nvidia lost $200B in market cap today. In the millennium, the word “online” was the bubble maker. Some will win, some are born to sing the blues.
Do you really need to ask Scott where to invest as a 65-year-old? Scott made his money founding companies. I retired at 56 because I foolishly but luckily put my entire 401k in tech stocks like MSFT and GOOG and NFLX and NOW during the tech stock bubble of the ZIRP 2010s. All my money is and all yours should be in cheap index funds like VOO. Scott said that last week on his podcast Pivot.
This newsletter came into my inbox just as I was using Claude to write a form letter for a non-profit’s volunteer onboarding process. AI has changed things – ever so slightly – for me. But as someone that has programmed 100’s of robots (including ones you ride at major theme parks) I continue to stand by my statement from a decade ago when confronted with the AI hype machine of the day:
“the robots aren’t coming ; they’ve been here for 40+ years. And – they continue to be really, really stupid”.
Claude did my menial writing task really well. But, I still had to proof and then contextualize its output.
And – it still cannot create a solution to the complex people-moving challenges we have in our community for bike lanes. I had to write a “tech note” about that (entirely written by me), with some real photos of the real problem and real photos of real solutions. And then share that with real human bureaucrats that still have a hard time wrapping their head around why they created things that often cause more harm than doing good… But I digress.
Glad to see you can also see the AI hype train Scott. It ain’t nothing, but it ain’t what we’re being told it is and the grifters and hucksters are having their time in the sun ; the dark winter is coming for these bad investments, and real people will have to do real work to cleanup the mess.
Godspeed.
Canadian Rocky
Spot on Canadian Rocky. In most cases — AI is an expediter, but you still need to understand the core drivers of business value (and how these are changing) to deliver results. AI won’t tell you how to do that. The Tempus example is interesting — as it really speaks more about a broader business opportunity to enable patient-self-management of their care. The AI used to create better lab tests are a simple G2M strategy for a broader play. Who cares that you leverage AI — tell me how you are leveraging new tech to truly address a big hairy problem that silo’d/financially strapped health systems cannot today solve? External PR should focus on the gains made with AI, and help folks understand how “tech for good” can make something possible that we all desperately need (consumers experiencing the healthcare they deserve).