Maybe the Fed should have already cut, earnings roundup, and more on AI data
📈 Trending Up: iPad sales … Threads … AI frenemies … HuggingFace competition … 3D image generation … trade tensions … link rot … drama at X …
📉 Trending Down: KOSA, in the House … Apple sales in China … Intel’s workforce … Knowde’s valuation … schools … the stock market …
📰 Good News: Evan is free.
Welcome to Cautious Optimism, a newsletter on tech, business, and power. Modestly upbeat.
Should the Fed have cut already?
The Fed met earlier this week and decided to leave rates where they were. Traders were not shocked, expecting the Fed to make its first rate cut in September.
Then Thursday came, with a sheaf of economic data that undershot expectations. Here’s how Yahoo Finance described the numbers:
Weekly jobless claims once again rose more than expected last week in the latest sign of a cooling labor market. New data from the Department of Labor showed 249,000 initial jobless claims were filed in the week ending July 27, up from 235,000 the week prior and the highest level since August 2023.
That led to some concern that the Fed may have left things too long, especially since other major central banks were getting busy with the knife. Stocks fell.
Enter today, with even more economic data that was a disappointment:
Nonfarm payrolls grew by just 114,000 for the month, down from the downwardly revised 179,000 in June and below the Dow Jones estimate for 185,000. The unemployment rate edged higher to 4.3%, its highest since October 2021.
Throw in slowing wage growth and the Fed has managed to cool inflation and a host of critical economic indicators with higher rates. Well done. Stocks are down yet again.
But hard landing or no, it seems that the economy has reached the point where the data that should trend up is trending down, and the data that should trend down is, in fact, mostly doing that.
The Fed could do a bit of catch-up by dropping rates by 50 bips (0.5%) in September instead of the minimum-viable rate cut of 25 bips (0.25%). We’ll have to see. But congrats to our central bank, for the national economy no longer appears to be overheated, or in danger of a wage-price spiral.
Earnings Roundup
Yesterday saw a massive grip of earnings data drop, so much that we’re going to have to be summary today. Here’s my video rundown on Amazon from TWiST, if that’s your jam, and here are video notes on DoorDash and Cloudflare.
To work:
Amazon: AWS profitability rose sharply as its growth reaccelerated. However, overall, Amazon's growth projections undershot street expectations, leading to a sharp selloff in Amazon’s stock today. I wonder if investors will ever demand that Amazon break into smaller pieces to unlock value because it really does seem that Amazon is probably worth a little bit less than the sum of its parts today.
Coinbase: A revenue beat and a profit miss were the name of the game in Coinbase’s Q2 numbers. The company’s ability to generate adjusted EBITDA in nearly any macroeconomic climate is to its credit. But Coinbase’s uneven top line results make it a bit hard to value the company. Stablecoin revenue at the company in the quarter set at least a local record, while employee count has found a new stable base in the 3,4000-3,5000 range.
Apple: Revenue growth of 5% to $85.8 billion led to Apple reporting 11% per-share profit growth to $1.40 in the second calendar quarter. Sales in China, and iPhone sales overall fell, while iPad and services revenues made up for the drops. Apple’s profit margin remains insanely impressive, even if its growth rate has become modest.
DoorDash: Investors were stoked that the food delivery company reported revenues of $2.63 instead of the $2.54 billion that was expected. Per-share losses were greater than expected, but with the company forecasting a slightly better Q3 GMV figure than the street anticipated, shares of DoorDash are eight points higher this morning.
Roku: A double beat in the trailing quarter was good news for the streaming concern, but shares of Roku are only up a point or two after reporting its second-quarter results. Why? Revenue guidance for Q3 was merely in line with expectations of just over $1.0 billion. That and I would hazard that advertising revenues could take a hit if the economy slows further.
Cloudflare: Which company is growing at 30% with improving gross margins and just beat both revenue and profit estimates for the second quarter? Cloudflare, whose shares are up just over 3% in early-morning trading today. Cloudflare’s results intimate that the most quickly growing software companies are not doomed to near-term deceleration into mediocrity. That’s encouraging.
Snap: Snap only has two moves on earnings day — sharply higher or sharply lower. Today is the latter, with the social media company’s shares falling around 20%. Why? CNBC notes that despite mostly meeting trailing expectations, Snap’s guidance was a bit light. Investors hate that, especially with the company reporting a return to both operating, and free cash flow negativity in the second quarter.
AI and copyright, part XVII
The RIAA is suing Suno and Udio for training their music-generating models on copyright-protected data. That’s old news. The latest update to the story is that the two companies have responded.
Both companies now admit that they ingested and used proprietary data to train their models. They argue that such use is fair use, while the RIAA does not agree. I wonder if we will see a settlement here or whether the RIAA intends to see the case through. (Put another way: does the RIAA want a pound of flesh or a decapitation?)
Regardless, the admission by the startups is clarifying, if not surprising. What else would they have trained their AI models on besides the world’s collected musical recordings? The RIAA’s argument that the two companies’ products are copyright-destroying machines that can ape existing tunes incredibly closely is not entirely answered by the startups’ position, in my view.
It seems odd that we’re seeing a scrap of this size continue when it seems that both sides of the suit simply need to cut a revenue-share agreement. The RIAA and its constituent members would get paid, and consumers would be able to create disposable musical creations that they send to friends (the main Suno/Udio use case that I have seen thus far).
The reason why that is not happening is greed. Not in the negative sense, I suppose, but just a desire to not pay for training data. We can know that thanks to Perplexity.AI, another AI product that has been dinged in the public sphere by folks who make data that companies like the startup use to train and feed their models.
After taking knocks Perplexity is working to share future ad revenue with publishing partners. Perplexity’s Sara Platnick told The Verge it’s an approach not popular with its own backers:
“By the way, our investors don’t love that we’re doing this because they’re like, ‘Oh, we want you to have the same margin profile as Google,’” Shevelenko said, adding that Perplexity can’t compete with Google by mimicking their strategies. Instead, he says, the company wants to focus on building a profitable business by forming alliances with the media by creating the right long-term structures, like ad revenue sharing.
Good on Perplexity, even if their moves feel slightly tardy. Suno, Udio, take note.