An update on my experiences with Ozempic, then thoughts about the AI competitive landscape and what Elon Musk believes.
Back in September, I started taking Ozempic both because I have Type 2 diabetes and because I’ve struggled to lose weight for a long time. I’ve been open about this, writing about it in this newsletter, and many readers of The Dispatch have kindly asked me how it’s going?
So far, so good.
My doctor wisely put me on a subclinical dose for a month in an attempt to dodge the well-known GI side effects, and it worked. I’ve had no intestinal pain, no diarrhea. Indeed, I used to have intermittent heartburn, and that has vanished. My blood sugar was better at my last test, and I’ll test it again in the next few days.
I don’t weigh myself, but people (including my doctor) say I’m losing weight. Apparently, my face looks different: it’s hard for me to see because I look at my face every day. My pants are getting baggy, and some shirts that used not to fit now fit again, which suggests that positive things are happening.
The differences that I do perceive are in my appetites. I’m now crazy for dried cranberries and baby carrots. Cottage cheese and dried cranberries is the breakfast of champions.
I still have a sweet tooth, but the kinds of sugar I want are different. I used to love gummy candies, particularly sour gummy candies. I was curious, so a couple weeks ago I bought a bag. They did nothing for me and actually made me a little queasy. But I love the miniature dark chocolate peanut butter cups from Trader Joe’s. Yum. I’ll eat between two and four of them, whereas before I might have worked through half a container.
I’ve never wanted breakfast first thing, so that’s no change. When working hard, I’ve often forgotten to eat lunch, so that’s not new. I eat less at dinner, at least I do if I haven’t skipped lunch.
One quarter into a year-long process, things have proceeded at a slow but steady pace.
The AI Biz Competition isn’t like earlier Big Tech
BrandRank.ai CEO Pete Blackshaw did a hands-on-keyboard “what can this thing do?” exercise on DeepSeek, the Chinese GenAI platform that is scaring other AI companies and Wall Street. He wrote up his experience and analysis for Advertising Age ($). Here’s one insightful snippet:
DeepSeek isn’t just another AI tool—it’s a glimpse into how the battle for attention and “answers” is evolving, and perhaps more importantly, a sign that the AI playing field is leveling.
This is worth digging into. AI Optimization will emerge as a big new part of marketing. (Pete’s company, BrandRank.ai, is a pioneer in this space.)
Unlike Search Marketing and Search Engine Optimization (SEO), which for the last 25 years has really meant “Google Marketing” and “Google Search Optimization,” there is no clear winner-takes-all company in the AI space, and it’s unlikely that there will ever be one.
What’s happening with AI companies is more like what happened in the early days of automobiles than what happened with search.
Between 1895 and WWI, there were 1900 car companies in the U.S. alone creating over 3000 makes. Then came consolidation down to Ford, Chrysler, and General Motors, the Big Three, who owned the U.S. market until imports started to gain big market share in the 1960s. These days, car companies tend to be global.
Something similar is likely to happen with AI services: proliferation then consolidation. A handful of years from now, there will be a lot of different AI services, assistants, and agents. This will commodify access to AI and drive prices down for end-users. Commodification will be a big problem for the AI companies themselves, which have been had valuations zooming into the stratosphere. Commodification will drive those valuations back down to Earth, which will lead to consolidation. Nvidia’s stock fluctuations after the success of DeepSeek are a telling example of more to come.
Musk Redux
A paragraph in an article about the business of AI in the January 25th issue of The Economist caught my eye:
Instead of domination by one firm, some expect model-making to be more like an oligopoly, with high barriers to entry but no stranglehold—or monopoly profits. For now, OpenAI is the leader, but one of its main rivals, Anthropic, is reportedly raising money at a $60bn valuation, and XAI, majority-owned by Elon Musk, is worth $45bn.
This supports what I was talking about with regard to AI proliferation and consolidation, but the thing that made my eyebrows perk up was the $45 billion valuation for XAI. That’s a familiar number: it’s very close to the $44B price tag that Musk paid for Twitter.
As longtime readers of The Dispatch will already know, for years I’ve contended that Musk’s rightward swing has nothing to do with his personal politics: he just wanted to make it thinkable for conservatives to buy Teslas. I still think that he wanted to get out of buying Twitter, but after the Delaware courts forced him to keep his word he decided to extract as much influence out of Twitter as he could.
Since Musk took Twitter private and changed it to X (ick), the company has lost over 70% of its value, according to an estimate by Fidelity. He doesn’t seem to care, and the XAI valuation is a hint about why that is.
Musk’s only politics is a love of money. He has no other beliefs. He has cozied up to Trump in order to have more influence on the government, reduce his own taxes, and deregulate the industries he dominates (electric cars, space flight, satellite internet). His malignant Department of Government Efficiency is in service of the same goals.
Musk is not a free speech absolutist; if he were then he wouldn’t silence his critics on X. I doubt he really even cares about “the woke mind virus” he complains about on X. If it turns out that he doesn’t really want to go to Mars, I won’t be surprised. I don’t believe anything he says about anything involving values, morality, or culture. He says those things simply to keep stirring things up so that, when he does want to exert influence, he has the attention of his 215M followers on X.
Musk is shameless: it is nearly impossible to embarrass him. The only person who seems to get under his skin (maybe) is his trans daughter Vivian, whose social media posts are worth reading.
That $45B valuation for XAI, though, is interesting. It’s unclear whether Grok—the GenAI agent created by XAI and accessible through X—used Twitter data as part of its training. Given the misinformation rampant on X, that would contribute to AI hallucinations. What is clear, though, is that the investors who supported Musk’s acquisition of Twitter got a big chunk of XAI, which will more than make them whole even if Twitter’s value declines to zero.
That only happens, though, if valuations stay high for AI companies, which is getting less likely. If Musk tries to unload XAI in the next 18 months either to an acquirer or a sudden IPO, that will be a hint that he agrees.
Note: If you’d like to receive articles like this one, plus a whole lot more, directly in your inbox, then please subscribe to my free weekly newsletter!
* Image Details: I took a selfie, and then used Adobe Photoshop to remove the background, make it black and white, and then used the Poster Edges filter to make it look like a comic book illustration.
Leave a Reply