Welcome to Cautious Optimism, a newsletter on tech, business, and power.
Happy Thursday! It’s GPT-5 day (more on that in a moment), but to kick off let’s talk about Smartsheet. Previously a public company, Smartsheet is a no-code automation and collaboration platform, and one that I used to chat with after earnings. I recall those chats because its erstwhile CEO, Mark Mader, was a persistent bear concerning inflation and the economy. Talking to him was always fun, because his take on the state of things was often orthogonal to what others had to say.
I missed it somehow, but Vista and Blackstone bought the company for $56.50 per share in January, after announcing the deal in late 2024. Selling your company for $8.4 billion is a big deal, of course, but what do you do afterwards? A half-year later, we found out: Mader is out of the company — both the CEO chair and the board — in September.
Why the quick exit? Having worked for a company owned by private equity, I had a hunch. So, off to GlassDoor CO went to see if we could hunt up any notes from staff regarding the new regime. As expected, we found reports from staffers irked at pricing tweaks impacting customer relationships, changes to comp, reports of offshoring, loss of equity compensation (expected), remote decision-making, and more. In short, it appears that Smartsheet is getting PE’d, and the former CEO probably didn’t want to be a faux lord of someone else’s fief. Just spitballing, but here’s hoping that the humans still at Smartsheet do alright. To work! — Alex
📈 Trending Up: Duolingo, after earnings … global trade, ex-US … Google’s need to defend itself … Perplexity-POTUS … PE interest in the datacenter game … AI carbon emissions … for now, at least … youth unemployment … Replit’s revenue growth …
📉 Trending Down: Enjoying nature … US-EU relations … Anthropic’s human churn … Apple’s AI team … global trade freedom, as of midnight … regulation-by-the-merits … Replit’s margins, at times … Intel stock, after POTUS called for its CEO to resign … shame …
Anyone remember calls for less Federal intervention in the economy? Trump 2.0 is using political ties to make antitrust choices, trying to boot public-company CEOs, forcing American tech shops to make less money through state-directed investments, and more, demanding POTUS-controlled golden shares in industrial companies, and taking direct stakes in minerals production. Wild.
Notes from the Earnings trenches
Airbnb: The company beat revenue and profit expectations in the second quarter. Like Uber, Airbnb has discovered that generating lots of cash is great, but it can also pile up. On the back of $1.0 billion worth of free cash flow in Q2 and $11.4 billion worth of “cash and cash equivalents, short-term investments, and restricted cash,” Airbnb intends to rebuy more of its equity. With just $1.5 billion remaining in its prior $6.0 billion buyback capacity, the short-term rental company is adding another $6.0 billion worth of share-buying firepower to its books. Investors, however, seemed more let down by future growth prospects than they were by seeing their relative portion of ownership increase. Shares of Airbnb are off nearly 8% this morning.
AppLovin: Shares of mobile app distro and monetization company AppLovin are mostly flat after reporting 77% revenue growth in Q2 to $1.26 billion. Net income, adjusted EBITDA, and revenue all hti all-time highs in the quarter. And thanks to buybacks, AppLovin has largely halted the expansion of its float. Why the tepid reaction to stellar results? Share of AppLovin have appreciated from the low $40s to start 2024 to just over $390 before earnings. High expectations mean outlier results can, at times, not be enough to delight investors.
DoorDash: Shares of American delivery giant DoorDash are up more than 7% this morning thanks to a revenue and profit beat from the former startup. Orders rose 19%, total platform spend grew 20%, and revenue for DoorDash ticked up by 24%. The company posted its largest GAAP net inccome and what appears to be record adjusted profit as well. Sure, we all whine about delivery fees, but it seems that despite the whining, we’re all still in the private-taxi-for-your-burrito game.
Amplitude: Beating revenue expectations while coming in light on EPS per some estimates, Amplitude is a company showing signs of reacceleration. After reporting 12% ARR and 10% revenue growth in Q1, Amplitude turned in 16% ARR growth (to $335 million) and 14% revenue growth ($83.3 million) in the second quarter. Mix in improving cash flow results, and the product analytics company appears to be in rude health. Adding $15 million worth of ARR in one quarter, the “highest net-new ARR in nearly three years” per CEO Spenser Skates, is a good sign. Shares appear flat this morning.
In venture news, Skates’ spouse Anne Lee Skates — previously a partner at a16z and an investor at Floodgate — launched a new venture capital firm called Parable yesterday. Parable will write checks of up to $10 million, and has already led five rounds before its announcement. More in her thread here.
It’s GPT-5 Day
Gird thy loins, for GPT-5 day is upon us.
Today at 10 AM Pacific time, OpenAI will debut its latest closed-source model. The release of GPT-5 comes on the heels of two generally well-received open-weight models from OpenAI, and after a new world model from Google and an updated Claude Opus model from Anthropic and a new open-source model from the Qwen (Alibaba) team. It’s a busy week.
But all those other announcements should pale in comparison to what Altman and Friends have cooked up, provided that the model is as good as hoped. Should we be so expectant?
Reuters reports from testers that the jump from GPT-3 to GPT-4 was larger than we’ll see with the leap to GPT-5, but that the new model family is impressive.
An archived — whoops! — announcement page that got taken down for the GPT-5 model family writes it serves up “major improvements in reasoning, code quality, and user experience [handling] complex coding tasks with minimal prompting, provides clear explanations, and introduces enhanced agentic capabilities, making it a powerful coding collaborator and intelligent assistant for all users.”
Access to GPT-5’s flavors will be tiered based on how much people pay. TestingCatalog writes that paid subscribers will get GPT-5 replete with “advanced reasoning,” with even more goodies for those paying for OpenAI’s most expensive plan. Free users will get vanilla GPT-5, the publication writes.
We’ll see. Recall that OpenAI could be hooking up a tender offer not at $300 billion, but $500 billion. I presume that the trade will hinge somewhat on how well — or not — GPT-5 compares to market expectations.
At this point we wait. I for one am glad that the entire world is not staring at me today.
How AI makes money
The Collison brothers of Stripe fame are busy, but John has enough time to podcast and he recently hosted Anthropic CEO Dario Amodei for a chat. A few choice excerpts are worth your time. From the official transcript, all quotes from Dario unless noted:
On current AI use cases, and where future revenue will spring:
I would say definitely the application that has grown the fastest, although it's very far from the only application, we have a wide range of them, is definitely coding. And my theory on why it's grown so fast, other than that we focused on coding and the models are good at coding, it's actually really a statement about societal diffusion, which is that if we look at today's AI models, I think in every area, there's a huge overhang in terms of what they could do compared to how they're actually being deployed today because there's some friction. […]
So, I guess a way to summarize it would be to say that code is out in the lead, but we see a long tail of quite a lot of other stuff, including some very, very significant use cases. I think code is maybe an early indicator, like a premonition of what's going to happen everywhere else. It's the same exponential, it's just faster. It's just happening faster.
On investors not initially believing in AI revenue scaling:
Dario: AI is strange in a number of ways. I think one of the ways it's strange is that because it's an exponential, we have a hard time calibrating exactly how big the business will be. So, we had the following experience. So, in 2023, I'd never raised money from institutional investors before, and so our revenue was zero at the beginning of 2023 because we had not released a product. So, I was putting together something and I'm like, "Oh, I think we can probably get $100 million of revenue in the first year." And this caused some investors to say, "This is crazy. This has never happened in the history of capitalism. You've lost all credibility with me."
John: "You're just making up numbers." Yeah.
Dario: "Goodbye. Goodbye." And then we actually did it. And then the next year, I was like, "Oh, well, I think we can go from $100 million to a billion." And actually, having done it the first time, it was a little bit less dismissed as crazy but still often dismissed as crazy. And then, we did it again. And this year, we're halfway through the year; we're, as you mentioned, well past $4 billion in revenue in logarithmic space, to add another order of magnitude.
On per-model profit versus burn:
There's two different ways you could describe what's happening in the model business right now. So, let's say in 2023, you train a model that costs $100 million, and then you deploy it in 2024, and it makes $200 million of revenue. Meanwhile, because of the scaling laws, in 2024, you also train a model that costs $1 billion. And then in 2025, you get $2 billion of revenue from that $1 billion, and you've spent $10 billion to train the model.
So, if you look in a conventional way at the profit and loss of the company, you've lost $100 million the first year, you've lost $800 million the second year, and you've lost $8 billion in the third year, so it looks like it's getting worse and worse. If you consider each model to be a company, the model that was trained in 2023 was profitable. You paid $100 million, and then it made $200 million of revenue. There's some cost to inference with the model, but let's just assume, in this cartoonish cartoon example, that even if you add those two up, you're kind of in a good state. So, if every model was a company, the model, in this example, is actually profitable.
What's going on is that at the same time as you're reaping the benefits from one company, you're founding another company that's much more expensive and requires much more upfront R&D investment. And so the way that it's going to shake out is this will keep going up until the numbers go very large and the models can't get larger, and then it'll be a large, very profitable business, or, at some point, the models will stop getting better, right? The march to AGI will be halted for some reason, and then perhaps it'll be some overhang. So, there'll be a one-time, "Oh man, we spent a lot of money and we didn't get anything for it." And then the business returns to whatever scale it was at.
Maybe another way to describe it is the usual pattern of venture-backed investment, which is that things cost a lot and then you start making it, is kind of happening over and over again in this field within the same companies. And so we're on the exponential now. At some point, we'll reach equilibrium. The only relevant questions are, at how large a scale do we reach equilibrium, and is there ever an overshoot?
Tomorrow we have some AI reports to read (Menlo, SignalFire), and GPT-5 reviews to consider. And more earnings. Hell of a week!