Hot take monthly: tech layoff apocalypse, millennials ruin everything, and AI killed the rock star coder
testing out a new format
Hey all! I’m testing a new format with the newsletter. For this month, I’m writing shorter takes on a few hot topics. I think society needs more hot takes, not fewer, so trying to up the quantity!
This month: looking at why tech hired and then fired so many people; millennials causing havoc in the housing market (again); and whether AI will really replace coders.
Send me a reply if you like the new format (or not). PS - got a nice little shoutout for my book from the Verge!
What do all those tech employees do?
Mark Zuckerberg has declared 2023 the “year of efficiency.” He’s living up to it, too, laying off 21,000 employees.
But taking a look at pre-pandemic 2019 headcounts, Meta will still be ~50% bigger even with all these layoffs. It’s almost as if in 2021 headcount at Big Tech companies reached an escape velocity where hiring went completely out of control.
Many of the layoffs won’t even take headcount back to trend. So what’s the point? Either they should be cutting to the bone, Elon Musk style, or it’s all for show. All these rich, entitled people were probably doing nothing anyway, right?
I think the 2021 bull market created a unique problem: the “net zero” employee. This is an employee whose impact is small - perhaps a 0.01% improvement in a product. But when working at 10 billion dollar scale, that’s enough to offset their salary. The arduous interviews in Big Tech ensure that the people there are smart enough to find that balance.
Most companies wouldn’t bother to hire them, but at more than $1m in revenue per employee - higher than any industry except the cyclical and brutal energy industry - it’s much easier to justify.
Source: Calcbench
Source: Kipfolio
Plus there’s no real cost, and it keeps talent away from competitors and they can be “stashed” for later. And Big Tech is in a unique situation - there’s never been an environment where such huge numbers of employees can’t be clearly aligned with revenues and costs.
Banks and car companies, the megacaps of the past, could clearly align the debt product or factory where people work with profitability. But Big Tech might have 5000 product and engineering people aligned to an app, and it’s not clear which parts of email or iOS really align with the bottom line. It’s a subjective call.
Unfortunately, investors got tired of paying for those extra employees in the form of dilution from stock compensation.
How’d we get here? Likely a mix of empire building, risk aversion (e.g., more headcount on a project is less risk), and lowered expectations at size. That means there’s no single fix either:
Layoffs in middle management are needed as it has a multiplied effect. Fewer middle managers means less likely to hire additional reports.
“Tone at the top” is needed from management to encourage risk taking
HR changes are needed to change promotion criteria, career tracks, and org structure to keep the average YOE from getting too high
Layoffs were just the culture shock needed to get the ball rolling. The irony is that if that’s true, it’s a great time to be bullish on Big Tech.
Further reading: The Big Tech Rebound Is Underway - by Alex Kantrowitz, some good tweets by Karthik Hariharan on headcount and bureaucracy
The Millennial housing dilemma that will define the 2020s
Pity the millennial. If they just hadn’t eaten so much avocado toast, they might have a home.
But we, the largest generation in history, might have the last laugh. House to rent ratios are at all time highs, making most feel like they have to swallow the pain of getting the last “adult” hurdle of a house.
Source: Has Housing Bottomed? | Pragmatic Capitalism
But do they? The market for multifamily housing increasingly looks like it might go into oversupply, putting pressure on rents.
Vacancy rates are gradually rising, just as an all-time record of apartments is about to be dumped on the market.
Source: Calculated Risk
This is coming just as the YIMBY movement is starting to get serious momentum (something something government being late…) in policy circles and government.
It seems the rent vs own ratio will become increasingly out of whack - either rent becoming relatively cheaper, as millennials pile into housing as they did in 2021. This situation has downstream effects: after years of fertility declines, it looks the early 2020s actually had a mini baby boom. If millennials choose to stay in apartments, that might not last long, leading to lower economic growth in the future…but also possibly having positive implications for struggling city budgets, increasingly bitter school battles, the ongoing inflation riddle, the climate and climate policy (gas taxes anybody?), and general policy priorities (public transport and childcare versus highways and bridges).
Or maybe…The movement to get rid of single family zoning might fuel a duplex/townhouse boom…just in time for when millennials don’t need them any more. And maybe that’s a good thing: in general, the less spend on rent, the more money we’ll have for things we enjoy.
Further reading: Housing: Don't Compare the Current Housing Boom to the Bubble and Bust, Moody's: Multifamily Demand "Softened notably over the past few quarters" , Will We Ever See Affordable Housing Prices Again? - A Wealth of Common Sense
AI and the doctor dilemma
There’s a saying about doctors - “no other discipline works as hard to end its profession.”
Well, move over MDs, because it seems coders are intent on putting themselves out of a job? Now that chatGPT and Bard can write basic code, the most in-demand job of the last decade suddenly looks to be disrupted.
But the fact that doctors have become better and better at curing disease has not put any doctors out of a job - in fact, it’s increased the demand for healthcare more than ever, by leading to a growing and aging population. Prescribing medication is only one part - and making a diagnosis in an increasingly complicated world is getting harder.
Coding is likely the same. For anything bigger than a web scraper, the architecture matters more than the code - code was already free for every use case on the internet through GitHub and StackExchange! Scalability, security, and configurability are major keys to architecture versus writing a simple “count to 3” program. As software systems grow, the complexity of making them talk to each other becomes exponential - like any network. Now there’s some evidence LLMs can do this, but it’s not set it and forget it - we need good coders to understand if the code and architecture LLMs are generating is any good or not.
Moreover, the most automatable jobs have already had costs wrung out. Fifteen years ago, I got the advice to avoid software engineering - because it was all going to be outsourced. And a lot has been - every US company has support people in Manila and IT people in Bangalore. But these are more complementary than substitutes to the best jobs. Ironically, these are the jobs most threatened, and so the returns to AI might be a lot smaller than we expect - initially.
Calculators didn’t reduce demand for math, it made math omnipresent. GPT will do similar things for code.
Faisal! You are as sharp as ever! Loved the quote .. if millennials didn't eat so much avocado toast .. haha