AI is in the news lately. Oh, you noticed? There’s been so much AI news that at times it has almost cut through the political chaos. 

There was the day last week when US stocks lost more than $1 trillion dollars in market cap because of a newish Chinese AI named DeepSeek. I’ll tell you more about DeepSeek below. Nvidia alone lost almost $600 billion in market cap, the largest loss in value in one day of any company ever in the history of money.

Or maybe you saw Trump’s inauguration day announcement of Stargate, which might or might not involve $500 billion of private investment in AI infrastructure in the United States by OpenAI, Softbank, and Oracle. It will create hundreds of thousands of jobs, honest, really, they pinky swear. Softbank is the Japanese company led by Masayoshi Son that lit $16 billion on fire in the disastrous WeWork fiasco, which the New York Times called “an implosion unlike any other in the history of start-ups.” Fun coincidence: Masayoshi Son attended Trump’s press conference about Stargate and it was just like his meeting with Trump eight years ago where he pledged to invest $50 billion in the US and create 50,000 jobs, except most of the Softbank money went to WeWork where it burned up, oopsie. But hey, maybe Stargate is totally real and it will happen even if Elon Musk thinks it’s a pipe dream, who knows, right?

Sigh.

AI is unavoidable even if you’re blocking out most of the headlines in 2025 for mental health. We’re constantly being pinged with updates about the latest AI-driven changes to our phones, our computers, our programs and apps.

It’s overwhelming and chaotic.

There isn’t any way for you to keep up. I’m going to encourage you to look away from news about AI. Paying close attention to daily news about AI is like trying to understand a thunderstorm by analyzing each raindrop.

If you’re not an AI developer or financial analyst, there are more important things to do than follow the breathless daily details of financial dancing by gazillion-dollar companies. Stick with the big picture.

In future articles I’ll try to convey the impact that AI will have on society (basically the reason all these resources are being thrown at it), and what you can do today to use AI and get a sense of its power. 

But I know, you’re curious about DeepSeek. I get it. 

Let’s go back in time to ancient history, roughly three years ago.

AI researchers did groundbreaking work and in 2022 OpenAI introduced ChatGPT, built on a new way of analyzing data. Details don’t matter, just assume that AIs work in a fundamentally different way under the hood from traditional computers and servers.

Many people saw the potential for AI to reshape our world in important ways, unlike cryptocurrency, a gambling fad which is useless today and will be useless tomorrow.

The major tech companies immediately began to pour massive amounts of money into development of AI infrastructure, more than $500 billion in 2024 and even larger amounts projected in 2025. There was a broad understanding that the future of AI development was tied to more server capacity and more powerful processors trained on ever larger data sets. Microsoft, Google, Meta, and others worked on the assumption that victory would go to the largest companies with the biggest server complexes. This shared belief drove the stock market growth of Nvidia, which makes the biggest chips to run in those data centers, until it became the most valuable company on earth.

When OpenAI publicly unveiled its first AI model in 2022, a debate immediately flared up about the destructive possibilities of advanced AI. That was combined with fear of China – everyone is terrified of China, right? The result was a ban on shipment of high end AI processors to China starting in 2022, in the hope of slowing down China’s progress in developing sophisticated AI models.

Have you heard about the law of unintended consequences? What happened next is an unintended consequence of the US chip ban.

Some Chinese companies, including DeepSeek, stockpiled less powerful AI chips, the ones that weren’t covered by the US chip ban. The Chinese companies didn’t have the money to build massive data centers, and they didn’t have powerful processors.

So they improvised. Chinese engineers worked out short cuts and hacks and clever workarounds, doing the best they could with little money and low-powered chips.

In January DeepSeek released a free model for testing that performs (almost) as well as the most advanced AI from the US tech companies.

This caused an uproar, because DeepSeek spent a tiny fraction of the amount that the US companies are spending to train their models. And the DeepSeek AI is optimized to work on the low-power chips that it had, since it couldn’t get more powerful chips.

In other words, in one fell swoop, DeepSeek challenged the business model of every major US tech company. 

And the world went crazy and the stock market did its short-attention-span thing.

Behind the scenes for the last three years, the giant tech companies have been engaged in a constantly changing blur of strategic pivots, shifting alliances, and investments in research and infrastructure.

Now Microsoft, Google, Meta, and the rest are adjusting their business strategies, changing their development efforts to incorporate the lessons taught by DeepSeek, focusing on efficiency and cost reduction, and trying to act as if everything is normal and they weren’t caught with their pants down.

The next three years are likely to be even more frenzied and harder to follow.

For 48 hours or so, DeepSeek seemed like a harbinger of doom for US tech companies – a sign that the future holds fewer data centers, less powerful chips, less energy consumption.

But there are good reasons to think the lessons learned from DeepSeek will only accelerate the spread of AI. Satya Nadella, CEO of Microsoft, put it this way:

“Jevons paradox strikes again! As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can’t get enough of.”

Not much has changed. AI is still revolutionary technology that will be embedded in almost every part of our world before long. We’ll talk about that next time.