So much for global trade
Welcome to Cautious Optimism, a newsletter on tech, business, and power.
Happy Friday! Stocks aren’t down 5% in pre-market trading, the sun is out here in Providence, and I am feeling better than yesterday. That means it’s going to be a kickass day. Hugs, and let’s have some fun! — Alex
📈 Trending Up: Emergencies … AI safety? … palace intrigue … Netflix stock, after 13% revenue growth in Q1 … consumer prices … Exaforce … university spinouts … Thesus … greentech M&A? …
The screaming in my head, after all the bluster it turns out that Trump had no plan, no leverage, and is ready to abandon Ukraine. Though, this could be good news of a sort?
📉 Trending Down: Public safety … national cybersecurity … the amount of information Netflix will release … middle-management at Intel … blanket tariffs … encryption? …
The global economy? The trucking market is soft, shipping rates are going wild, and export-dependent industry is about to collapse per Flexport’s founder. But don’t worry, rising fees on Chinese ships will sort it all out.
Jason and I hosted the founder of a clothing company on TWiST earlier this week, if you want a more visceral look at what even Trump supporters are going through in the current trade war.
Google, the monopoly
Meta is currently in the dock over its alleged monopoly in the social media space. Rewinding the clock to 2020 when the suit kicked off, the FTC wrote that Meta,
not content with attracting and retaining users through competition on the merits, Facebook has maintained its monopoly position by buying up companies that present competitive threats and by imposing restrictive policies that unjustifiably hinder actual or potential rivals that Facebook does not or cannot acquire.
For a more recent slate of arguments, you can check out the government’s opening statement slides from earlier in the week. (Facebook rebranded to Meta back in 2021, mind, so the language in the above is dated.) Notes on Zuckerberg’s testimony can be found here, if you want even more.
Put all that to one side. Meta is not the only big tech company in the United States facing the potential for its business to be reordered by government dictate. Recall that Google was found last year to have monopoly power in search (the fight in that case is now over remedies).
But Google is in even deeper water. The company was also taken to court over its advertising business, which it was alleged to hold monopoly power over “three digital advertising technology markets.” The judge ruled as follows:
With the benefit of a three-week bench trial and extensive post-trial filings, the Court finds that Plaintiffs have failed to prove that there is a relevant market for open-web display advertiser ad networks, but have proven that Google has violated Section 2 of the Sherman Act by willfully acquiring and maintaining monopoly powering the open-web display publisher ad server market and the open-web display ad exchange market, and has unlawfully tied its publisher ad server (DFP) and ad exchange (AdX) in violation of Sections 1 and 2 of the Sherman Act.
Google, naturally, tried to spin the above as a win:
We won half of this case and we will appeal the other half. The Court found that our advertiser tools and our acquisitions, such as DoubleClick, don’t harm competition.
We’ll see. (More on the split decision here.)
Despite the White House swapping out its key resident, the push to rein in the market power of some of the largest domestic technology companies continues apace. Nothing’s done until it’s done, but with Meta at risk of being forced to divest Instagram, and Google potentially forced to divest Chrome and open up its search data in addition to whatever remedies might pertain to the advertising challenge — we could be looking at a very different technology world once the legal dust settles.
A fun question to ask yourself: Will the domestic technology market be more competitive (good) or less competitive (bad) if Alphabet and Meta are forced to sell key assets?
Instagram on its own would be worth a mint, yeah? As an indie entity, it could go after the company formerly known as Meta. And that would necessitate competing over talent (acquihires) and product (acquisitions). If we’re all about capital recycling to keep the Silicon Valley flywheel spinning, how much sweat do folks want to spill over defending incumbents?
Startups > BigTech
Quote of the Day, via the FT:
The ups and downs of AI land
While the world is busy sorting out freshly barricaded trade corridors, the AI game is as hot as it has ever been.
New models from OpenAI this week cracked the top spots of leaderboards, with folks pointing out in response that Google’s Gemini 2.5 model family offers nearly as good performance for a fraction of the cost. No worries, OpenAI quickly rejoined, you can use its o3 and o4-mini reasoning models for half the cost so long as you are willing to wait for the processing to occur.
And that’s not all. Google is offering free premium AI access to college students to try to get them hooked on its AI ecosystem.
We’re watching a Mag7 company fight tooth and nail and fanged GPU with an up-and-comer (comparatively), leading to not only rapid technological improvement but also falling effective prices. Great news for the market, great news for startups building atop AI models from the two companies.
You love to see it, yeah?
There’s a cost to speed, however. In early April, TechCrunch reported that Google was working to release models more quickly, even if their safety cards weren’t yet completed. A later release of more information from the search giant didn’t calm the waters. OpenAI has been similarly dinged for moving quickly with less regard to safety lately — and in the past. But the company is also not doing nothing, either.
I stand on the side of AI companies should move fast and we can sort out the risks later camp much more than the we should slow down now and get things right before making more intelligent machines side of the fence. For two reasons:
I do not view incremental AI model improvement as a risk to human safety
Slowing domestic companies will only cede the AI market to companies built offshore
You may disagree, of course (and if you do, hit reply and tell me why?), but the current market balance in AI-land is speed and power over safety and caution. Probably for the best, but worth our attention all the same.
Cracks in open-source AI
Closing, Meta is not infinitely wealthy, it turns out. Here’s The Information on a pitch from the social giant to other tech companies looking for help paying for its AI cost basis:
Meta Platforms over the past year asked Microsoft, Amazon and others to help pay the costs of training Meta’s flagship large language model, Llama […] the news that Meta tried to tap other companies to help pay for Llama shows how the mounting investment required for AI is beginning to stress one of the richest tech companies in the world.
Meta, still setting billions of dollars alight on its VR play, is also spending heavily on AI compute and model development — models that it largely gives away while also using internally on its own products. What’s weird about Meta and its AI strategy is that it lacks normal monetization channels:
Building and supporting AI models makes sense for companies that offer compute to other companies, as the work can drive training and inference workloads on their offered clouds.
Building and supporting AI models makes sense for companies that offer paid access to their work, bringing in incremental revenue as usage rises.
Meta, in contrast, lets folks use its AI models in whatever cloud or computing setup they wish. That’s great for developers — if Meta makes killer models that are free, then AI costs diminish, right? — but perhaps less so for Meta proper. For the open-source approach to work out financially for Meta, it needs to do one of two things:
Earn enough from internal usage of its AI work to subsidize its public-benefit releases (open-source-ish Llama models)
Start to sell access to its servers and offering a hosted Llama experience at scale that recoups its investment
Otherwise, Meta is just spending to stay near the top of the pack while not earning as much as its rivals. That’s an expensive tactic to take on the market, especially as OpenAI accumulates new paid subscribers and hyperscalers keep, well, printing cash.
We used to hate this
Tae Kim, a journalist and author of The Nvidia Way — which you should read; it’s great — tweeted out a few lines from the Declaration of Independence today that caught my eye. I verified the excerpt via the National Archives.
Here are some of the stated issues that led to the American Revolution:
