Hey folks. A bunch of marketing-adjacent businesses and tech are experiencing successive shake-ups these days. We’re about to take a look at one such case and begin with the fact that a Chinese ChatGPT knockoff took Nvidia out of world’s-most-valuable-company contention since our last blog post.
Nvidia doesn’t sell oil, phones, ads, or a continuous blast of adolescent brain-wrecking vertical videos. It doesn’t have world class logistics like Amazon. What Nvidia did well was surf two Silicon Valley hype-waves in a row. It served as the backbones for crypto mining and generative AI processing, seamlessly transitioning from one stock-pumping buzzword to the next without tumbling when Wall Street swapped them. That’s not nothing, and Nvidia does develop and manufacture sophisticated hardware, but the double-bubble party looks like it’s now over. Watch this space though.
We’re talking about a graphics card maker in a marketing newsletter because we’re staffed with power users of both ChatGPT and Claude for copy and coding, both of which heavily rely on Nvidia’s wares. DeepSeek, a Chinese LLM that was inexpensively developed and has incomparably lower operating costs, came out of nowhere this week and shamed the best that OpenAI and Anthropic (the organization behind Claude) have to offer.
You can take a look at the screenshot above this paragraph to appreciate the marketing implications of DeepSeek. It’s currently the most downloaded app for iOS and Android in the United States, France, Canada, Australia, China, and the UK, and is the second-most downloaded app in Germany, Japan, and Italy (three countries that we haven’t seen grouped together in a while).
DeepSeek is popular elsewhere, but all of the countries we just listed are top ten global economies (except for Australia, which is still rich). i.e. where people who are likely to pay for a $200 monthly OpenAI subscription live. It’s where the people who pay $25/month live too. The same goes for $20 Claude subscriptions. It’s hard to find a benchmark where DeepSeek doesn’t outperform those LLMs, and, again, it’s (currently) free.
You can see how the OpenAI CEO coped with DeepSeek in that screenshot (plus the anime-profile-pic guy who nearly ratioed him). The CEO of Anthropic wants to “beat China with self-improving AI” and called for stricter export controls on US chips going to China. Also, breaking news– OpenAI just released o3-mini, a new model that is “93% cheaper” to run and can reason better than its previous flagship model. We received an email about it as we began typing this paragraph so we have no experience with it yet, but we can link you to OpenAI’s o-3 mini prompting guide if any of you want to look into it.
The GBP post we asked DeepSeek to compose is one of a thousand use cases for LLMs in the digital marketing space, which is why we don’t feel bad focusing on them for an entire newsletter. Google Search continues to play catch up with AI outfits to its detriment (look at the results Google serves up for a nonexistent movie), but of course remains the hub that facilitates everything in the local search ecosystem. In any case, we’re putting it in timeout today.
We’ll circle back to the big G next week. We’re going to get into our closing links now. The first of which is also digital marketer-adjacent, especially for any of you who’ve ever worked from a coffee shop: Starbucks bathrooms are now for paying customers only. Also, to follow-up on one of last week’s links, some leaked documents mean that we now know Meta’s handing out $300k contracts to TikTokers for Reels-exclusive content. Third, from Buffer, is a study showing that consistent social media posting leads to more likes and engagement. Finally, from Later, we’ve got a guide for brands to help them determine if and when influencer marketing is worth paying for.
That’s all for now. Take a look at this change your password day guide (it doesn’t load quickly, but it has reasonably good tips) if that’s an occasion you plan to observe tomorrow. Take care.