Amazon had a lot of revenue in Q2. And they’re on track to have a lot this quarter too. Imagine that, right?
Big-cap US tech earnings wound up on Thursday, when the brainchild of civilian astronaut Jeff Bezos said sales were $134.38 billion last quarter. That was easily ahead of estimates, and well above the mid-point of the company’s guide from April.
Top-line growth is back in the double-digits, and it’ll stay that way based on the Q3 guide.
Amazon said sales this quarter should be between $138 billion and $143 billion. The midpoint is $2 billion better than consensus expected.
Andy Jassy called it “another strong quarter.” I’ll save you the trouble of parsing the release: Everything beat, although that comes with the obligatory caveat that the market reaction hinges on how investors interpret the AWS results and any AWS color from the call. It’s the same dynamic you see with Microsoft and Azure.
AWS sales beat for Q2, rising 12% in the process. Jassy said growth for his brainchild has “stabilized.” AWS customers “started shifting from cost optimization to new workload deployment,” he remarked. I’d be inclined to call that music to investors’ ears but, again, it’s difficult to know how the market will trade the AWS color.
Naturally under the circumstances, Jassy talked up “a slew of generative A.I. releases ” in the cloud. Amazon is making it “more cost-effective for companies to train and run models.”
Operating margins at Amazon were 220bps better than expected. Operating income more than doubled from the same period last year to $7.7 billion. That was $3 billion ahead of estimates.
Meanwhile, Apple reported a third straight YoY decline in revenue, as expected, but $81.8 billion was a slight beat. Consensus was looking for $81.6 billion.
That’s the longest streak of YoY revenue declines since 2016.
iPhone sales of $39.67 billion were short of estimates. Sales of wearables and iPads also missed, coming in at $8.284 billion ($8.38 billion expected) and $5.91 billion ($6.33 billion seen), respectively. Mac revenue beat by half a billion.
Services revenue of $21.21 billion came in stronger than forecasts, as did Greater China sales of $15.76 billion. EPS of $1.26 beat, gross margin was better than estimates and the company’s cash pile, at $28.4 billion, was $4 billion larger than analysts saw.
As ever, it’s hard to be “bearish” or downbeat about Apple’s prospects just because analysts wanted a rounding error’s worth of additional iPhone sales last quarter. Apple has some new products coming up, and eventually I assume those goggles will be slimmed down, and cheaper such that they become a serious addition to the lineup a few years out.
Services revenue notched a June-quarter record this year, the company has more than a billion paid subscriptions and Luca Maestri noted that the installed base of active devices hit a new record “in every geographic segment.” Again, it’s hard to be overtly bearish. It’s Apple.
Tim Cook offered the usual feel-good assessment. “From education to the environment, we are continuing to advance our values, while championing innovation that enriches the lives of our customers and leaves the world better than we found it.”
Chuckle as you will. But can you say the same thing about anything you’ve built? I’m not sure I can, much as I like to think I’m “enriching lives” in these hallowed pages every day. Certainly, I’ve got a lot of work to do if I want to “leave the world better than I found it.”
AMZN report was v good, esp margins, AWS growth (+12% v cons +10%/feared +SD%), NorAm rev +11% (cons +7%, consumer holding up), advertising rev +22% (far above GOOG/META). Guidance intriguing in a good way; hard to get mid-pt of rev guide without AWS ok or better (my opinion). Jassy talking the desired talk: efficiency/margins, lower capex, AWS stabilizing, AI AI AI.
AAPL report looked meh. iPhone miss, Services beat, Mac/iPad down hard, Wearables flat, GM% up from Services boost. Guiding Sep qtr rev YOY similar to Jun qtr’s -1.4%, FX headwind 200 bp less – so constant currency growth decelerates? Working the segments vs Sep qtr rev and GM% guide, it is hard to see iPhone upside and easy to see downside (my opinion).
H, I recently saw the movie “Oppenheimer”- which I highly recommend.
Having never met you and only knowing what you have chosen to tell us; to me, you seem more akin to Oppenheimer than Heisenberg. Oppenheimer was not only an intellectual who was capable of building a nuclear bomb; but he was also philosophical about himself and what he had done.
Now see Breaking Bad. Heisenberg features rather more prominently in that.
That was “H”- version 1.0.
Glad you’re not leaving this world soon as you are making the place a better place than you found it!
The “more cost effective to train and run models” is an interesting one! As part of that I think he means bc you can spin up resource intensive VMs as needed and terminate them or stop them when you’re done. But, at least for GPU based workloads I am finding it harder and harder to get available GPU VMs, and I have been wondering about what will happen with that. Maybe companies will start to buy a bunch of reserved ones upfront for years. Or maybe some will just decide to leave a subset of static ones up all the time. It can be pretty frustrating to get the “no available capacity” errors! I have been wondering about how that will turn out especially as I see such a huge influx in cloud based gpu node demand on my side. You would think that will start to impact earnings and customer cloud provider preference.
Thanks. That real life Intel is very useful.
I’m wondering how many firms (Actual users/customers!) really need to run LLMs on huge databases including 25 years of weather data in the Andes and attendance at AA Baseball games in the US over the same period when they simply hope to fine tune their automated call center applications.
Now we are starting to read more about how edge AI on chips from Intel and eventually AMD may make much more sense than loading up on racks of $35,000 NVDA GPUs designed to run LLMs.
Whodathunkit?
From a trading perspective I always wonder about the numbers getting calculated and how the team decides if it’s worth the cost of generating them. For example, for Trading Team A if their dashboard doesn’t update that morning with newer versions of those numbers in some way it impacts their trading decisions negatively. But, I don’t know how they assess that quantitatively (despite the fact that they’re all insanely good at math!). Which is to say I don’t know how they assess how much less money they’d make trading that day compared to the costs of running so many compute intensive processes to generate the newer version of the numbers that informed their trading decisions