DeepSeek Exposes Silicon Valley's Dark Underbelly

Its arrival has jolted the system. The Tech Crown is now contested.

Just days after its launch, DeepSeek China's AI chatbot became the most downloaded app on iPhone. As news spread that this chatbot offers high quality at a much lower price, the stock market reacted brutally. Nvidia plummeted. Alphabet, Amazon, and Microsoft quickly joined the rout, erasing over a trillion dollars in value. Companies within the AI infrastructure chain faced severe losses, while those building applications atop the AI infrastructure saw a slight boost.

Wait—what? How did that happen?

DeepSeek has matched the performance of leading American chatbots while using significantly less computing power, resulting in substantial cost reductions. This breakthrough suggests that the initial expenses of training AI models could drop sharply, lowering the costs of developing AI-driven applications and benefiting companies like Apple. With reduced computational demands, the market valuations of semiconductor manufacturers like NVIDIA and cloud computing giants such as Amazon, Microsoft, and Google have been adjusted.

Contrary to Silicon Valley's traditional view that AI training requires vast computing resources and heavy financial investments, DeepSeek has demonstrated high-quality outputs at a fraction of the usual cost. This shift has implications for companies like NVIDIA, Amazon, Microsoft, and Google, which previously saw significant cash inflows based on the old paradigm.

Cost Comparison between ChatGPT and DeepSeek.

The tech bros didn't complain as the money poured in. The efficiency question was largely ignored. When someone finally raised it to Sam Altman, CEO of OpenAI, during a VC conference in India, he was politely dismissive.

Question: How is a small, smart team with a budget of $10 million could build something substantial within AI?

Sam Altman: It's totally hopeless to compete with us on training foundationels.

Oh, well. This is precisely the kind of narrative Silicon Valley passed to its VCs: It's expensive to build, very expensive. But the long-term payoff, due to it being the latest tech revolution, will be worth it. I agree that AI is transforming the search industry and massively expanding our ability to process data and derive solutions.

Yes, American tech companies are innovating and challenging the current technological landscape. But are they challenging themselves on the efficiency front? The answer is a resounding and brutal no. This lack of focus on cost-efficiency is why tech stocks, backed by so much value, were routed—not because growth assumptions were shattered, but because cost assumptions were.

And this is a massive freaking problem for Silicon Valley. Throwing more money at the issue doesn't create better products; it creates bubbles, and bubbles burst. The Valley is now so engrossed in greed that the focus has shifted from revolution to making more money.

There is no better example than Tesla in this regard. The company triggered a massive, inside-out revolution in an automotive industry that had refused to change, stuck without even incremental innovation. Tesla shifted the paradigm, with the United States leading the charge in the tech-driven EV revolution, reshaping the global automotive industry.

Year after year, Tesla's EVs topped sales charts, signaling a shift where American companies could have dominated the global automotive market. However, as the company grew and its owner amassed wealth, his priorities shifted.

After years on top of the global electric vehicle market, Tesla has lost the title of world’s largest EV producer to Chinese powerhouse BYD. In 2024, Tesla made 1,774,442 electric vehicles, 4,500 shy of BYD’s 1,777,965.

Tesla squandered its advantage. With its early lead in battery technology and strategic supercharging stations, it should have never lost its crown. Similarly, other American car companies had what was needed to further penetrate the global automotive market but failed to capitalize on their opportunities.

The global passenger car market generates a staggering $2.29 trillion in revenue. The United States possessed all the necessary resources to dominate this market, or at the very least, significantly expand its market share. However, this advantage was squandered.

China now leads in EV production and sales, and we're seeing a similar issue with AI. The Valley is so focused on money that it refuses to consider efficiency, becoming greedy enough to chase the next shiny object once profits are secured.

If this trend continues, the next Microsoft, Amazon, Tesla, Facebook, or Google won't be American; they’ll likely arise in China, India, or Europe—anywhere but the Valley.

Consider TikTok, the latest social media craze. It wasn’t built in America, was it? All the Valley could do was get a bipartisan agreeement to ban it. Is this how Silicon Valley intends to maintain its edge—by banning competition?

Greed kills.

So, in many ways, I am extremely pleased that DeepSeek shocked and awed the tech world. I am even more delighted that the valuations of companies in the AI value chain received a brutal adjustment.

The warning for the Valley has arrived.

Now…. change.