Let's talk tech Thursday #5
Welcome back to the newsletter that lasts longer than 11 minutes and doesn't cost the earth.
In this week's edition:
- With AI-giant NVIDIA announcing a series of US-based factories, I muse on how much longer American companies will dominate AI,
- A new AI startup has me musing on whether the market is ready to explore the "how" of AI,
- Meta goes head to head with the Federal Trade Commission, and - yup, more musing - this time on whether or not this is the end for big-tech's hold over our data (I won't spoil it for you).
All that, plus: some backpedaling on electronics tariffs in the US, a rethink on who counts as a political ally from the EU, and a podcast recommendation on the ethics of technology in our long-term future.
Let's dig in.
Top Stories
NVIDIA to Manufacture American-Made AI Supercomputers in US for First Time
Summary
NVIDIA, the company behind a huge percentage of the world's AI hardware, is building factories to manufacture AI supercomputers and chips in the United States for the first time. These Arizona and Texas-based plants are expected to create "hundreds of thousands of jobs" and strengthen the American economy.
So what?
As Trump's tariff war seems never-ending - against China, at any rate - people have been left wondering where their stuff is going to get made. Until now, there hasn't been an apparent ramp up in plans to create US domestic manufacturing. NVIDIA look set to change this within the next two years.
There are a lot of questions still unanswered though, including "who is going to work in these factories?", "how much will they get paid?", and perhaps most pertinent for the world at large, "how much more expensive are my chips going to be?" It seems almost impossible that a move to US-based manufacturing won't dramatically increase the cost of the very building blocks of our AI infrastructures. And while NVIDIA and it's ilk still remain among the wealthiest companies on the planet, you can bet they aren't going to eat all of that cost increase.
If all of that is the case, we could see a real shift away from the current major players in the space - OpenAI, Anthropic, Microsoft, Google, etc - towards their Chinese counterparts. Deepseek made global headlines in January when it announced its R1 model and accompanying chatbot, in large part because it could get comparable results to ChatGPT on older, less powerful hardware. And Chinese AI hasn't slowed down.
While the US incumbents are going to be relying on their considerable head-start, and the fact that organisations have already embedded a lot of their systems with specific AI-products, there are almost certainly more than a few companies that are going to be looking at the price/performance ratio of AI over the coming months. The more agile of them might reconsider whether they can justify any increase in costs.
A lot of "experts" are going to warn people away from doing business with Chinese tech companies. But with 1.6 billion people on TikTok (and it being the world's most downloaded app of 2024), Deepseek wiping a billion dollars of the US economy overnight, and the recent wave of love for Chinese factories who are not only lifting the lid on luxury brands but offering the chance for consumers to cut out the middleman, I don't know how much the average person is going to care about any of that. If costs increase, are we sure that the demand for American AI is that inelastic?
I think there's one more thing to consider here. While the NVIDIA blog post doesn't address this directly, many of these factories are already wholly or partly autonomous. This high level of automation means that the labour cost of the factories isn't as high as you might expect. There are also vast improvements in "digital twin technology" - nerd-speak for "we can build the physical factory in the US, but have it operated remotely by people in any country we choose". All of this is me saying: maybe the cost increase won't be as high as expected.
"But Will," I hear you say "the article says that this will create 'hundreds of thousands of jobs'. Where are these if not in the factories?" While that is true, the figure is "over the coming decades" - a suitably vague and long-term timeframe. On top of that, my particular brand of cynicism reads that section not as "these factories will hire hundreds of thousands of people" but rather "the existence of more NVIDIA AI chips will mean more jobs". I think we have to wait a little more to understand the full impact of this.
Related story: Apple India Produces $22 Billion of iPhones in Shift From China
The next OpenAI: 5 things to know about Mira Murati’s Thinking Machines Lab and its $2B round ambition
Summary
Mira Murati, former CTO of OpenAI, has launched an AI startup called Thinking Machines Lab. They've recently gained two new advisors - Bob McGrew and Alec Radford, both also ex-OpenAI - and have doubled their funding ambitions to $2 billion without yet having a product.
So what?
If you believe the aims on the Thinking Machines Lab website, the company wants to make AI tools that are not only suited to individuals unique needs, but that are "more widely understood" than their competitors. If you're a glass half full kind of person, this reads very much like a company looking to crack open the black box of Generative Artificial Intelligence.
If Thinking Machines Lab successfully raise their $2 billion, and the consensus seems to be that Murati will succeed, it will be a strong signal that the market is increasingly more interested in the "how" of AI, as well as the "what".
It could be an exciting turn of events for consumer-AI, as individuals are questioning more and more what happens to their data. Mere days ago, ChatGPT announced that premium users in parts of the world would benefit from the model remembering "all your past conversations". This hasn't gone down well with everyone though, and some are concerned that this crosses lines they thought had already been drawn.
You might also be interested in: ChatGPT’s memory can now reference all past conversations, not just what you tell it to
Meta and the FTC face off in court over monopoly claims
Summary
The Federal Trade Commission (FTC) is arguing in court that Meta, the company behind Facebook and Instagram, has acted as a monopoly by buying competitors instead of competing fairly. Meta's lawyers claim that the company is not a monopoly since its apps are free and face competition. The outcome of this trial could lead to significant changes for Meta, including possibly breaking up its business.
So what?
Mark Zuckerberg's Meta occupies spots 1, 3 and 4 of the top 5 most popular social media platforms on the planet (Facebook, Instagram, and WhatsApp respectively, with YouTube and TikTok rounding out the list). Some have speculated that he may be the most influential person in the history of modern communications.
Meta's power comes from the vast amount of data it owns. At it's core, it's a digital marketing company. The amalgamation of data that sits across Facebook, Instagram, WhatsApp, (and Threads too, I guess) means it can lead the charge on hyper targeted ads in a world fueled by capitalism. As far as companies being able to print money, Meta are up there. Where it faces competition from scrappy start ups, it's usual M.O. is just to buy them.
If forced to break up the empire, Meta may well lose their edge. The immediate impact for you might be fewer creepily specific Instagram ads. Longer term though, it might open the door for a more competitive landscape for internet startups, greater diversity of social spaces, and even better online policies as platforms have to fight for user trust, not battle inertia.
More widely, success by the FTC would set a precedent that might be followed by other courts around the world. There are no shortage of potential monopoly and antitrust lawsuits. Perhaps this might embolden other governments to take a stand. At a macro level, we might well be seeing the beginning of the end of big tech's dominance over our data.
I'll not hold my breath for any of it though...
What else is going on out there?
Trump spares smartphones, computers, other electronics from China tariffs
Listen, I don't even know if this will still be true by the time you read this, such is the pace at which the US tariff situation is moving. But for a period of time (that we may or may not still be in), the Trump administration published a list of goods that were excluded from import taxes. This list was broad, but included some hardware and components, such as laptops, hard drives, and flat panel displays. It remains to be seen whether this will have an impact on how countries tariff the US, but for now at least consumers might be able to breathe a small sigh of relief.
EU issues US-bound staff with burner phones over spying fears
The European Commission is giving burner phones and basic laptops to some staff traveling to the US for IMF and World Bank meetings, due to concerns about data being stolen by the US government. This already happens in trips to Ukraine and China, for fear of surveillance by Russian or Chinese powers, and marks a very clear deterioration in EU relations with the US under Donald Trump's presidency. Non-paywalled coverage here.
Bizarre and Dangerous Utopian Ideology Has Quietly Taken Hold of Tech World
If you've got 40 minutes to listen to a super-nerdy but incredibly fascinating podcast about the long-term thinking behind some of today's tech trends, you could do worse than this one. Philosopher and historian Émile P. Torres talks to host Kelly Hayes about “a set of ethical views concerned with protecting and improving the long-run future.” While that all sounds very nice, Torres argues that this hypothetical good comes at the dramatic expense of very real people in the here and now. There's a transcript of the episode in the link, if you're more of a reader than a listener.
That's it for another week, thanks so much for reading.
There's never a shortage of technology stories, so I try to bring you stories that you might not necessarily read elsewhere. I'm always open to feedback though!
If there are types of stories you want to see more of, drop me an email and let me know.
In the meantime, I hope you enjoy a nice long weekend, and I'll catch you next Thursday.
Best,
Will