The first company with no employees wins
Welcome to Cautious Optimism, a newsletter on tech, business, and power.
Good morning! Hinge Health priced at the top-end of its range last night. Let’s see how it trades, yeah? Stocks are largely down again, home sales were crap in April, but manufacturing data came in stronger than expected. To work! — Alex
📈 Trending Up: BYD … the national debt … Musk on the Hill … antisemitism … violence in the West Bank … tokenized stocks … Vercel’s AI chops …
📉 Trending Down: Solar stocks … American support for democracies … short-seller profits … Russia … Russia … stocks …
Why the OpenAI-Jony Ive deal could work out
OpenAI’s decision to pay billions of dollars in shares for Jony Ive’s consulting-cum-hardware-cum-design-cum-AI company is a bit of a puzzler. You can render the huge price tag reasonable in a few ways:
OpenAI is spending about 2% of its total ~market cap (private-market valuation) on the deal, which means it only needs to be so accretive to the AI giant’s overall worth to pay for itself.
OpenAI’s design skills are slack enough that overpaying for name-brand talent is simply a smart way to lock up said talent, and prevent, say, Anthropic from pulling off the same move.
Or, that OpenAI simply wants more headlines to help support its expensive, private-market price tag.
Jason and I mostly focused on the first option on the podcast yesterday, and I think it’s a good lens through which to view the transaction. But since that chat, we’ve learned a bit more:
The FT reports that OpenAI already owned about a quarter of Ive’s Io concern, making the buyout worth about $5 billion instead of the full $6.5 billion figure that was reported.
The WSJ reports that Sam gave OpenAI staff a “preview Wednesday of the devices he is developing to build with the former Apple designer Jony Ive, laying out plans to ship 100 million AI ‘companions’ that he hopes will become a part of everyday life.”
The deal is, therefore, far more a hardware gambit than anything else. So, do we like the deal? For Ive and his shareholders, of course. But AI hardware has thus far proved a magnetic iceberg to startups looking to chart new courses, so you might want to take the other side of the bet.
Still, for 2% of OpenAI, Sam has bought himself a single at-bat in the AI hardware game. And a home run there could be worth a lot. Just ask Apple shareholders about the value of Cupertino’s iPhone business.
The transaction feels costly. But as the poker kids say, no gamble, no future.
OpenAI also announced a huge AI infra project with the UAE today, part of the US’s cozying ties with the autocratic petrostate.
The first company with no employees wins
It’s time to take a fresh look at the FRED chart tracking the number of software development jobs currently advertised on the employment board. The data shows if domestic employers have more or fewer dev jobs open than in prior years, starting in 2020:
From left to right, you can see a COVID-19-induced collapse, followed by the ZIRP era and a quick decline into mid-2023. Over the last two years, the decline in total software dev job postings in the United States has slowed, but the trend is still negative.
From a value of 100 set on the first day of February 2020, the COVID-low on the chart was around the 64 mark. Today? It’s just over 62. That means that the number of dev jobs on Indeed today is worse than during the panic layoffs the pandemic brought on.
That’s brutal for recent grads — something we discussed here — but also indicative of the vibes in tech-land. The push to do more with less, burn cash more judiciously (startups), and return cash to shareholders more generously (public companies) is paramount to other concerns.
You can think of the market today as the inverse of the talent-hoarding we saw back in 2021; hiring ahead of need is backwards in a market where individual productivity expectations are rising — for both financial and technology-related reasons.
But no matter how much you think AI is reducing the demand for human talent compared to tech companies simply realizing that they can demand more from their staff than before without risk, both are likely impacting hiring today.
In the wake of Microsoft’s recent decision to jettison 6,000 staff, or around 3% of its staff — including a host of developers — it appears that Meta also wants to get busy with the knife. Again. Here’s BI:
[Meta is] telling managers to put more employees in its "below expectations" tier, the lowest performance bucket, during this year's midyear performance reviews, according to a memo […] For teams of 150 or more, Meta wants managers to put 15% to 20% of employees in the bottom bucket compared with 12% to 15% last year.
BI goes on to note that back in 2022 Meta boosted those percentage-targets of its staff labeled as “underperformers” from a range of 7% to 12%, to “up to 16.5% of staff.” Those changes were later followed up by cuts of so-called underperformers. I presume the same thing will happen later this year.
Meta reported $42.3 billion in Q1 2025 (+16%), leading to $17.6 billion worth of operating profit (+27%), and $16.6 billion worth of net income (+35%).
Perhaps Meta’s hiring practices have been so lax over the years that it hired a bunch of duds that it is now flushing out (unlikely). Or the company simply needs fewer people than it once did, thanks to AI-derived programming assistants (probably at least partially accurate). Most likely, however, I think the company is confident it can simply wring more work out of fewer bodies by axing headcount and that the newly-burdened workers won’t quit.
Something, something, Archimedes.
Personally, if I were as rich as Meta and had spent all that money and time hiring so many smart people, I would have kept them. But that’s why I will never, ever be a CEO. I am far too soft.
Where’s the above heading? We’re not going back to a developer paradise, when available talent was thin on the ground, and companies voracious to spend. At least for years to come. So we’re in a new world, and comments like the following from Replit CEO Amjad Masad earlier this week at a Semafor event (I interviewed Amjad for TWiST interview in March, if you want even more):
Masad said startups at Y Combinator are vibe coding their products with tools like Replit. Founders told him that while they thought they would need a chief technology officer, they turned to Replit first to see how much of their product they could code without a software developer. They said, “We’re on month three and haven’t had to hire anyone,” Masad recounted. “We think of Replit as our CTO.”
“I don’t think we’re there yet, where they can run the entire company without hiring engineers, but that might be a year, 18 months away,” Masad said.
That’s not going to make developers more in demand. Anecdotally, I am hearing from more and more founders that I interview that they have pretty modest hiring plans. Gone are the days of scaling up senior leadership and filling new floors with engineers, turning newly sold shares into employee 401k contributions and rent payments on the double.
The situation, while winsome for investors — less burn at portfolio companies means more durable investments, so long as growth rates remain hot — is not great for junior developers looking to get first experience. Who are not cracked as the saying goes. Or yet, I suppose.
Earnings rundown, Snowflake Edition
Shares of data company Snowflake are up 10% today, after it reported earnings yesterday following the bell. Revenue was up 26% to $1.04 billion, with net retention at 124% and 34% RPO growth to $6.7 billion. The company was comically unprofitable on a GAAP basis (-$429.7 million in the quarter), but if you yoink share-based comp costs and a few other items, the company’s non-GAAP net income was a tidy $87.5 million.
What mattered to investors was that Snowflake’s trailing results were better than expected, and its product revenue guide came in above expectations.
Frankly, Snowflake doing well in a period of time when data is maximally valuable and created in ever-larger quantities is not a surprise.
On the AI front, the company said that after just “getting started” a year ago, it now has “over 5,200 accounts using our AI and machine learning on a weekly basis.”
Much ado about Perplexity
Yesterday, CO wrote that Perplexity’s reportedly modest net revenues in 2024 must have greatly accelerated for it to be on the cusp of landing another nine-figure round at a newly heightened valuation.
A sharp-eyed reader took exception with the take, essentially arguing that we were being too generous. Applying the ever-useful idiot razor — not that the simplest explanation is the best, but that the stupidest possibility is most likely correct — we could infer that investors are simply overpaying for present day ARR at the company.
Perhaps. But I thought that everyone learned that paying triple-digit multiples for ARR was a bad idea?
Bueller? Bueller?