Google Announces Fourth Quarter and Fiscal Year 2023 Results [pdf]
abc.xyz
abc.xyz
- 2022: 190,234
- 2023: 182,502
Total increase in revenue of 10 billion.
Not sure what conclusion to draw from this, but being a software engineer I found employment count interesting. Though 5% change in head count seems hardly major.
the lesson here, i think is that there are limits to how much can be done, without moving into new industries and spreading horizontally.
They'll probably grow back to 190k within a few months, but those 8k new hires will probably take months if not a year to fully ramp up.
I feel like this is actually a good way of doing things, and wish Google did this more broadly. Just think if everyone who would have been laid off was instead transferred to a project that needs their skills, or given training to move into a role that's needed. Can you imagine what type of employee loyalty and morale that would engender?
Values shifted with the focus on earnings and growth getting less impressive.
Imagine how obnoxious code reviews would be if everyone wanted to submit as many lines as possible for their salary to go up, that is how management headcount requests works today.
Middle managers serve a role but honestly many of them can be replaced without really disrupting anything. At a certain point, you become too far removed from the actual work that is being done to really matter.
On the other hand, there are many low level managers with teams of fewer than 10 people who can be very valuable and hard to replace. People with deep expertise in the specific area they work on.
Measuring a manager by their headcount is just a lazy way to evaluate impact and calculate salary.
Because all the high value stuff(high ROI stuff) is already done or already being worked on. all that's left is low ROI projects which take up huge amounts of time but bring little value. employees and managers know they generate little value, so they end up doing more of them to compensate which makes the problem even worse as there's less and less high value stuff to do.
there's just too few opportunities, hence the need to expand into new industries.
i sometimes wonder if they would have more success with atoms based innovation rather than bits (reference to avc article on the matter). but that would necessitate hiring something other than software engineers.
Yes. However how do you find those less useful positions? And identify whether it's the developer who is less productive or the position?
For top management the easiest way is to fire people broadly and assign it all over equally by pushing the decision down the management chain. Medium managers know their teams a tiny bit (but often not much ...) than too management and identify the least troublesome positions.
And then you let them bring up cases to keep their headcount or hiring new.
Works, after a short bump in productivity, if it's just (human) resources and numbers in a sheet. (Similar to other budget cuts, where you give a Mandate to reduce cost by X% everywhere and then see who cries the loudest and who adapts) Of course has impact on trust etc. which has an invisible cost.
In fact it's even worse than that. A lot of Google's employees are dedicated to maintaining tools and infrastructure that Google built up over decades and hundreds of thousands of engineer-years because "the cloud" didn't exist when Google started. These other companies get all that "for free" (or rather, for nominal pay-as-you-go fees -- but certainly without the massive engineering investment that Google put in over the years) by building on AWS, GCP, or Azure. Without public clouds and the vast growth in open-source cloud infrastructure tooling, those other companies likely wouldn't be viable businesses.
Not contradicting your point really, just a note
Extremely cool individual who I miss. Maybe I should say hi to him.
I suspect that many good jobs went to people who learned about them in some way other than official job postings. (This is generally true.)
My favorite here is Bridgewater and their HR/Evaluation/Alignment system. There's books written about it. Developing it and running it took a high percentage of workers and contractors. Its implementation had all kinds of negative effects on everyone else, including the people doing the actual investment. But it took a pandemic, and the CEO being sidelined, for the whole thing to be basically thrown away.
Pandemic and larger-than-life CEO stepping down both seem like the major outliers in the dynamic you described. Bridgewater was very, very successful for quite some time with that policy in place.
I get it, developing and supporting some feature at a company like Google requires more people (both in engineering and elsewhere) than if you banged out a prototype of something similar at a startup. But I've read stories where hundreds or thousands of employees were laid off, and they were all working on some small service (or even an obscure feature of that service) that never had a chance of moving the needle for the company and offsetting their costs.
I'm sure most of those employees were hard-working, motivated, and tried to build the best product possible, but there are no other companies on earth that would pay that many people top-of-market salaries to work on anything that isn't existential to the company. Other companies would simply go out of business if they operated this way, but in big tech companies, this is lost in the noise when times are good. But when things slow down, big tech companies start to pay very close attention to this.
I guess we have different friends, but that doesn't match my experience. Of course, pumping yourself up and making yourself look more impressive for perf/promo is a huge thing, and accounts for a lot of pointless stuff being done just so you can say you "launched"/"landed" something. But the flip side of that is the recurrent memes about stuff like so-and-so over in that other division solved some incredibly hard technical problem, and "I'm just here copying protos around." Lots of Googlers realize they are just worker drones.
2019: 118,899
2020: 135,301
2021: 156,500
2022: 190,234
2023: 182,502
https://www.macrotrends.net/stocks/charts/GOOG/alphabet/numb...
So could there still be 10-20% extra as workers?
- 2021: 156,500
2023: What?!?! Laying off a fraction of recent hires? The world is ending!!
Obviously this sucks for the laid-off, but…you don’t need a very long memory to put this in perspective.
Yes, some tenured people suck. But it’s their salary that drives this behavior.
2021: this is clearly a bubble
2023: execs do "shocked Pikachu", "we have hired for a different economical environment!"
[0] https://www.spglobal.com/marketintelligence/en/news-insights...
Of course, the counter argument is that technology allows Google to do more with less people. I am not convinced by this argument, because Google's competition has about the same ability to do more with less people. If you can improve individual productivity than why not scale that across more people.
To summarize, this indicates to me that Google is reducing their value on human capital.
Say that made 3 revenue products revisions slip by 2 quarters, and not ship six cost-saving optimizations (if you make google search 0.001% faster, you save millions in compute). Further, say that made 100 companies not renew their 7-figure ad contracts. If all of this was evenly spread throughout the year, it still hits revenue and profit only a comparitive small amount in one year.
It takes time for those sorts of changes to propagate all the way through the system until they finally impact revenue and costs. The effects are long term.
I'm on record as saying that Google hired way too much and way too fast, but it isn't quite as simple as, X-people produce Y-revenue, so (X-1)people produce Y-Z revenue.
I mean that in several ways: Google doesn't have nearly as good a record of successfully recruiting (or retaining) top-tier talent as they did just 5-10 years ago. It is no longer seen as a dream position, or even a strong candidate's first choice. Similarly, hiring managers who see Google on a resume don't treat it as highly as they did 5-10 years ago.
Google is rapidly on its way to becoming just another big boring technology company, like SAP or Salesforce or Oracle. Now the culture of those around it has finally started to catch up.
https://seekingalpha.com/news/4060311-alphabet-falls-4-as-ad...
The post I was replying to was implying that cutting employees boosted revenues somehow.
Google Bets lost $863M in the same period, much less than ever before.
"Other Bets, which includes the Waymo self-driving car business and the Verily life sciences unit, reported revenue of $657 million, up from $226 million the year prior. Its loss narrowed to $863 million from $1.24 billion."
So revenue up 400 million, and losses dropped by 400 million.
The core concept is being proven pretty well, in markets with good weather and good roads at least... whether they can achieve some sort of critical mass that allows them to recoup any of the investment, even on a single capitalized vehicle... who knows. I've heard each Jag costs ~$250k fully loaded... that's a lot of ~$14-on-average trips before breakeven, and obviously the trips aren't free.
That all being said, riding in one feels like the actual future, and if they close in on this first-mover advantage and keep on keeping on, and strengthen their Uber integration, and open in more and more markets... there's a real chance they establish themselves and make this thing actually capture a significant portion of the trip economy.
That being said, it's not generalized, so opening markets is expensive because they're all individually trained for each mile of road in the market, I expect they'll be functionally useless in bad winters (but that's not a dealbreaker, per se), and those operating expenses remain high. Though I do expect we'll begin to see those lower, now that they're so deep into the proof of the concept. (They're finally starting highway driving here in PHX, which is really the last hill to climb.)
They said it costs as much as a “moderately equipped S-Class” 3-4 years ago. At the time, that would be ~$140k. That includes the base car cost, sensor hardware, compute, redundant vehicle systems and their integration. So not quite $250k, but still steep. I think the Waymo-Geely robotaxis coming next year will be interesting cost-wise.
> That being said, it's not generalized, so opening markets is expensive because they're all individually trained for each mile of road in the market
This is a misconception. The Waymo Driver is, in fact, designed to be generalizable. There some steps like mapping and initial validation in each market, but the same software runs on Waymo vehicles in every city. It’s definitely not individually trained for each mile in each market.
Oh, that's very interesting. Being in the test/training market, and having read that they were training on every mile while they were here, I thought that was a key differentiator, but if it _is_ generalizable, that's absolutely huge.
(And living here, during training, we'd see trains of 5-6 Waymos cruising down the streets of our neighborhood, so it kind of tracked with what I'd already heard from their videos and posts.)
Thanks for the correction. It's amazing what they've accomplished, and it's also a blast to show out-of-towners who have no idea this is _widely available_ here in Phoenix.
I switched to more or less only using Waymo for ride-share in SF after I got into the beta last year, and almost without exception every Uber or Lyft ride I have had to take since then for timing, availability, or route reasons has been some kind of debacle. Ranging from just an unclean and odd-smelling car to a ride my partner and I had to end early with the "pull over now" button when the clearly intoxicated driver missed multiple exits and slid on and off the highway median figuring out what to do.
By the way, that driver was a "Platinum" tier driver per the Uber app. I reported what happened in detail in a message to support, specifically using the safety issue area of the app, and got back an automated message explaining to me that my credit card charge differed from the amount shown onscreen because the ride ended early. I have previously ranted at length on HN about the nasty dark patterns Lyft uses to attempt to trick you into using their "priority pickup", which is always slower than promised and condescends to you with fake "just a few more seconds" messages while it waits for a driver. Recently, my Uber app has begun auto-selecting Black cars for no conceivable reason, except that it does this after 10pm on the weekends when it's more likely I'm in a rush and/or sleepy and/or intoxicated.
The user experience of Uber and Lyft has gotten so poor it feels like outright hostility from these apps towards the users. I've certainly had multiple drivers who behaved outright hostile to their customers. By comparison, the Waymo experience is perfect.
Experiencing Waymo has also altered my standards. Now I really notice how often the car is dirty and unkempt, stinks of pot or cigarette smoke, or the tire light is on, or the driver is handling the car poorly. Waymo is usually the same price, and none of that ever happens. If they pull this off I don't see the business model for human based rideshare surviving unless the cost of AV rides skyrockets once they need to be independently commercially viable.
I also live right off the highway, so I think I am more prone to getting people who pick up my ride on the way to or from somewhere else and then get upset about where they have to go. Pretty much the only place I need an Uber to these days is over the bridge to the East Bay when BART is not realistic, which is usually because I'm in a rush. One of my recent trips I wound up late anyway, because the first driver who pulled into my complex explained to me "because the uber app gives me these bad rides" I should cancel the ride for him, and pay a fee, because it was so unreasonable for me to expect him to deliver me to the destination I had called the ride for.
Edit: on another bridge trip, my driver was excited to show me his 'tiktok famous' friend on his dash mounted iPad while we went over the bridge. Diamond driver, perfect rating, surfing social media with one hand and steering with the other. Maybe Waymo has just made me picky and most rideshare users now expect to share use of an iPad with the driver in traffic
On the other hand though, I have seen misplaced items in Waymos a couple times and when I reported them in the app they pulled the car for cleaning ASAP. I actually did leave a water bottle once - a disposable, single use one - and they emailed me asking me if I wanted them to return it! That kind of vehicle management feels more like a benefit inherent to centralized taxi fleets than a benefit inherent to AVs, but it's something Waymo can take advantage of. I would guess the in-car cameras would be capable of detecting vomit-related events etc too.
I didn't know they had those. Wonder if they'll eventually add a mode where you could opt-in to livestream games like Cash Cab or Carpool Karaoke. Would make the commutes more interesting :D
Google expanded testing in Buffalo, NY in November 2023, including new sensor hardware: https://www.reddit.com/r/SelfDrivingCars/comments/1aedrmy/ne...
It makes sense to start with easier markets to a certain extent...though SF is definitely not an easy market in terms of urban design/clutter. It would've been way easier for them to follow up Phoenix with Vegas.
If they can handle SF, then aside from weather they can probably handle just about any other city in the US.
But for it to work, that entity will have to have ultimate control over where the vehicle goes. Everybody is going to want to have that ultimate control, and not cede it to someone else. That would make licensing not really work out since Google would retain control.
If letting your rideshare company decide your destination was really a thing, why wouldn't Uber be doing it already be doing it with human drivers?
> why wouldn't Uber be doing it already be doing it
They are already most of the way there [1]. Driverless vehicles will be cheaper to operate which will make physical locations more amenable to paying to acquire visitors.
Not yet...
> If letting your rideshare company decide your destination was really a thing, why wouldn't Uber be doing it already be doing it with human drivers?
Because Uber isn't an advertising company. They're still chasing fares like scrubs
My math says that's about 25 trips per day to break even in two years (modulo maintenance costs).
So you can make a pretty big business just pretending winter doesn't exist.
Cruise attempted to just start launching before the tech was really finished and it backfired horribly. Tesla is still doing Elon things, looks like the tech is mostly spinning its wheels as they work on yet another hardware revision that'll definitely handle self-driving this time.
Public companies can structure their business however they prefer. If the story needs to be that Cloud is growing and profitable on a numbers sheet, you need no more than the finance geeks to make it happen.
Market beating returns: Alpha. Long bets – Bet. Alphabet.
People in cloud are profoundly unhappy. It's scary.
I don't see any likely path by which Kurian would become CEO of Alphabet.
A little bit more Amazon/Oracle, a little less traditional 'Google'. I think most people work 45 hours a week. I don't see anyone slacking off. But it's not a death march, it pays well, there are tons of perks still.
shrug It's not some magical fantasy land that some people make google out to be, but it's a nice job with interesting problems, smart coworkers, and it pays well.
You don't understand why people fear the culture changing to something else? Most people there like the Google work culture, of course they would fear the culture becoming more like Microsoft or Oracle.
When I worked at Google I had no deadlines, full control over what I do during my work days, full ability to prioritize different projects or help people get things done if they ask etc. If that suddenly changed it would feel horrible, as you can understand, if you are used to a typical work culture maybe you don't see it but if you are used to the typical Google one they at least had before then it is a massive owngrade.
However, Cloud seems pretty intent on shipping and meeting deadlines so if we commit, I do feel pressure to make that deadline.
- Revenue increased from $76B to $86B (13% increase)
- Operating income increased $18.1B to $23.7B
- Net income increased $13.6B to $20.7B
- Diluted EPS increased $1.05 to $1.64Having read the book, the section "Change in Useful Lives of Our Server and Network Equipment" immediately caught my eye. It's a classic maneuver.
Amazon has multiple books with that title.
I've seen Google blow away earnings and all expectations and be down 10%, and completely miss expectations and pop up 5%.
Here, GOOG was $144.99 on Jan 18 and $144.50 on Jan 30 and this is supposed to be some dramatic shift in things? Bullshit.
[0] https://www.kotaksecurities.com/articles/it-is-priced-in-wha...
[1] https://finance.yahoo.com/news/alphabet-misses-expectations-...
[2] https://www.investors.com/market-trend/stock-market-today/do...
https://finance.yahoo.com/news/stock-market-today-nasdaq-fal...
Revenue, excluding traffic acquisition costs: $72.32 billion vs. $70.97 billion expected ($63.12 billion in Q4 2022)
Adjusted earnings per share: $1.64 vs. $1.59 expected ($1.05 in Q4 2022)
Cloud revenue: $9.19 billion vs. $8.95 billion expected ($7.32 billion in Q4 2022)
Ad revenue: $65.5 billion vs. $65.8 billion expected ($59.04 billion in Q4 2022)
I hope that 0.3 is not the reason for that slip haha
By the way people are asking chatgpt or copilot what they used to Google. Not everything but a lot. I most certainly do and I think I'm not alone.
https://www.bloomberg.com/news/articles/2024-01-18/microsoft...
I know for sure some people who use chatgpt in place of Google search like students or people needing translations.
Right, they got $300 million less than expected.
Which means they only got 99.544072948% of what the analysts expected.
That means they missed analyst expectations by 0.455927052%.
Not for total Revenue. Not for EPS. Not for net income. Not for diluted EPS. For the one category, Ad Revenue.
Resulting in a 6% after-market drop.
0.455927052% miss. 6% drop.
I think it's entirely possible that people focus too much on short term metrics and not enough on long-term growth.
The company gives guidance, so the estimates are not pulled out of thin air. Google should know their ad business very well at this point.
People selling after an ad miss is medium/long term thinking. They could be wrong, but was the miss caused by a wider economic slowdown (see UPS earnings) or was it Google specific? Either way, it could be sign of Google slowing down or the economy as whole.
That's not true, and hasn't been for a decade. For this earnings release, the proportion of revenue from ads was 76%.
> The company gives guidance
No, they don't. As far as I know they've literally never given guidance.
Also, if you only consider the search revenue you'll err in the opposite direction. It's 55% of their revenue, so pretty soon not even a normal majority. You should look at the "Google Advertising" line.
If you truly believe Google releases guidance, how about you just find a quote for what that supposed guidance was for Q4? You won't find it, because Google does not release guidance. The GP just made it up, just like they made up the numbers.
GCP is growing to be a meaningful part of Google's revenue, but the actual profit at Google is almost entirely ad revenue. If that is a miss, and on the horizon we see AI disrupting search and increased competition in digital video advertising then what does that mean for the future of Google?
I think that's probably being a bit pessimistic, but that's probably the argument.
Besides, it’s a good opportunity. If it’s down, the stocks are on sale. I still regret not buying Meta when the markets were crashing last time.
Every buyer and seller has their own price and own reason for it.
1. "sell on the news"
2. results vs "expectations"[1]
"expectations" is a weird Wall St concept where some small number of people who don't necessarily know much get to invent a set of numbers for the company results in the future. Then a bunch of other people who know even less believe/pretend that those expectations mean something, and will buy/sell the stock accordingly when the actual results are revealed. Yes Kabuki Theater. Actually it's about as bogus as software project planning/management, so perhaps all fields have similar made up nonsense. Hopefully not the people designing bridges. Or airplanes. Oh wait...
Total costs increased from $57.9B to $62.6B, and Net income increased from $13.6B to $20.7B
But the numbers appear to be billion (no way revunue 80mil) in the following table? what am i missing
More and more people now realize that the ad/attention economy is overall a bad thing for internet/the world.
Googler who at some point were seen as above-average engineers are now seen as leetcoders TC-Maximizer with very little ethical thoughts or opinions on what they work on.
All of those changes are a positive thing.