Earning Those Earnings 📧

Today starts a string of tech earnings with Google (and Snap, per below). Tomorrow is Microsoft and Meta. And Thursday is Apple. Going forward, the key thing to listen for in all of these earnings calls will be talk about how AI is impacting both the top and bottom lines. My bet remains that at some point, Wall Street will turn against the spend much as they did with the content spend by all the entertainment players a couple years back.

Back then, they boosted the streaming services for gaining users during the pandemic, then hit them for that spend once the world started to return to normal. Here, the AI affiliation has boosted all of these stocks, so the question is what will turn the sentiment? The simple realization that the spend isn't growing revenue fast enough? Profit? Hurting the focus on the core businesses? Something else? Or will the technology prove a boon to something like advertising faster than anticipated? NVIDIA doesn't report until November 20, so everyone has a little bit of time not to panic leading up to that report...

Apple's earnings seem particularly well-timed given their three products-in-three days cadence that will wrap on Wednesday. And the fact that Apple Intelligence – finally – went live in its initial, simple state yesterday. But as always with their earnings, all that will really matter is the iPhone and if their marketing team has done enough to convince people to upgrade to the iPhone 16 models even though there was no AI at launch. There will only be about a month of that data in that regard, but any comments about next quarter – the holiday quarter – will be key.

One more thing: one of Apple's next announcements – the new (tiny!) Mac mini – appears to have leaked early via Amazon. They've had some problems in that regard recently, it seems...

Update: And sure enough, here's the new Mac mini – and the 'Ternus Note'.

Subscribe to Spyglass

Email sent! Check your inbox to complete your signup.

I Think…

🍎 The 'Ternus Note' Is Back – While I'm generally not a fan of Apple's over-baked pre-recorded keynotes these days, I remain a fan of doing these videos in lieu of (or alongside) press releases for new products. And alongside the new M4 iMac announcement yesterday, Apple brought back this concept which shall henceforth be known as the 'Ternus Note' since it always seems to be SVP Jon Ternus (he who is perhaps next-in-line for the Apple throne) who MCs them. They're good. They're fast. And unlike when compared to a live event, where such things feel too slick and lacking energy, these are much better than a standard press release. I think Apple should do both: Ternus Notes for smaller updates. Live keynotes for bigger ones. [YouTube]

📰 The Hard Truth: Americans Don’t Trust the News Media — Jeff Bezos has responded to the blowback about his decision to end endorsements in The Washington Post. If we take him at his word (and no reason we shouldn’t, despite his poor judgement here), it sounds like the situation was very similar to my benefit-of-the-doubt scenario laid out a few days ago. In that post, I made the case for why he needed to respond to this directly; but honestly, this isn’t enough. Written a year ago, this op-ed would have been fine, perhaps even good — written on October 28, 2024, with a presidential election in a week, it’s weak. Framing this as “inadequate planning” skirts around the real issue: was the decision intentional, despite realizing how this would be perceived? Or was it a failure to recognize what the reaction would be? It’s not clear which is worse. Someone who can’t read the room or someone who doesn’t want to read the room. He talks about “sighing” when he realized the optics of the Blue Origin/Trump meeting on the day of this announcement, so he cares about how things look, yet he killed the endorsement anyway. Despite the fact that letting it run would have not been remarked upon, but ending the practice, well, had the opposite effect. Which his editors tried to convey. He didn’t understand, or didn’t care. Neither a great quality in a person talking up trust. That’s the hard truth. [WaPo]

👻 Evan Spiegel Battles to Revive Snap’s Fortunes – This feels like a particularly important earnings for Snap later today given both Spiegel's promises to focus the business more but also for the AR dreams that he so eloquently showcased last month. Company control aside, he simply won't have permission from Wall Street to do the latter if he can't prove out the former. The FT details how that tension is going over internally, but it also reads like it may have boiled over already. Hard decisions have been made, now it's time to execute to get that market cap something higher than a mere 1% (!) of Meta's. [FT 🔒]

🔎 Meta Develops AI Search Engine to Lessen Reliance on Google, Microsoft – Yeah this makes sense since Google and Microsoft have an obvious natural advantage in what is undoubtedly a significant subset of AI queries because they have real-time search results from the web via their search engines. So Meta can pay for such data, as it now does, but those terms are only going to get worse. This also speaks to a rationale for OpenAI's search product, which will obviously not fully displace Google but could yield results in this way, quite literally. As always, all eyes will shift to Apple here. If/when the default Google search deal is struck down by the courts, are they at all enticed to work on their own search product? It's not about straight web search necessarily anymore, but about having that information on the ready for AI... [Information 🔒]

🦙 We Finally Have an ‘Official’ Definition for Open Source AI – It's almost as if the OSI looked at Llama and said, 'yeah, our definition is going to be the opposite of that'. That's awkward because Meta is one of the partners in the initiative and helping to fund the project. Obviously, they're not happy with this definition, instead suggesting that things need to be less black and white in such a fluid environment with nascent technology. Also, they play the safety card – which is interesting since that's the one often used against them when they push new models of Llama out into the world as their main competitors keep everything much more on lockdown. It's worth noting that Mistral also doesn't fit with the OSI's definition – they're closer than Meta, but still have commercial usage restrictions that don't meet this new definition. [TechCrunch]

I Wrote…

Mark Zuckerberg’s “Thoughts on Open Source AI”

A page out of Steve Jobs’ playbook to go after... Apple (and everyone else)

SpyglassM.G. Siegler

Is AI Spend the New Streaming Spend?

Wall Street loves to reward arms race spending -- until they don’t.

SpyglassM.G. Siegler

I Link...

• Peter Cushing is being digitally resurrected – again! Star Wars fans will recall when Grand Moff Tarkin was brought back for Rogue One in 2016 – an early version of the now hotly debated "deep fake". While that usage is set to go to trial, this one pushes on to celebrate the 90th anniversary of Hammer Films (which made many Dracula and Frankenstein films in which Cushing starred before his death 30 years ago). [Deadline]
• I'm shocked – shocked – that Google and Microsoft are still "shadow" lobbying against one another after all these years. [CNBC]
• While Apple and Jon Stewart may have not mixed, clearly he still does with Comedy Central as the two sides have agreed to keep running it back on The Daily Show (once a week, with others hosting on other days) beyond next week's election and through at least 2025. As he quips, "I was really hoping they’d allow me to do every other Monday, but I’ll just have to suck it up." [CNN]
• Meta pushed out their own version of Google's NotebookLM in the form of NotebookLlama (lol) – it's "open", but not by the OSI definition (per above), but mainly, it seems like it's not as good. [TechCrunch]
• While the US got the first iteration of Apple Intelligence for iPhone and iPad yesterday and many other countries are getting it in December, the EU will start to get it in April – a delay, of course, due to the DMA. Apple undoubtedly wants to ensure they won't get fined for rolling it out, deeply integrated, as they are in other countries. As such don't be shocked if some features don't make the cut. There's always the Mac though, which has no restrictions including with Apple Intelligence because the EC didn't deem it a "gatekeeper" platform. [MacRumors]
• What/Who is this new "Red Panda" AI image generating model that is providing great outputs 100 times faster than DALL-E? [TechCrunch]
• Why did Jeff Bezos need to respond to the presidential endorsement fiasco? Because over 200,000 people had cancelled their Washington Post subscriptions as a result of the maneuver, reports NPR. That's nearly 10% of the entire subscriber base. That's wild. [NPR]
  • Meanwhile, the LA Times saw at least 7,000 cancellations due to their own pull back from endorsing a candidate, but they also only have around 400,000 subscribers to begin with (and it may have been higher, this is just the number they can directly tie to the move). [LAT]
    • Update: the numbers are now more like 250,000 for WaPo and 10,000 to 18,000 for LAT. [NPR]
• With Evan Gershkovich now safely home, the deeper reporting about what happened can begin. While Gershkovich clearly wanted the assignment, was it reckless for the WSJ to allow him to be put in such a situation in Russia? This will all just build until the release of his eventual book on the matter. [Semafor]
• Samsung TV Plus, their free, ad-supported streaming service baked into their TVs, is roughly as large as Tubi, Pluto, and the Roku Channel – around 80M - 90M active users. If anything, it's probably a good case that all the FAST players should have/build their own TV hardware and vice versa... [THR]
• While their partner Microsoft tries to get some sort of consumer AI traction – is that working, at all? – OpenAI is happy to flex their strength here: 75% of their revenue comes from consumers paying for their service. [Bloomberg 🔒]

I Quote...

"So far, AI is a crap poet."

-- Margaret Atwood, telling Reuters, amongst other things, that she's "too old to get too worried" about AI. "But if I were 30, I'd be worried." [via TechCrunch]