Jaws of Disruption: DeepSeek and the AI Shockwave

“If you want to talk about a perfect movie, you talk about Jaws.”– Quentin Tarantino

I have always loved momentum unwinds in the market. Making money should never be easy, and being long a crowded story that has gathered more rental investors like a snowball rolling down the hill getting bigger with each turn until we are reminded that the market is always humbling. This week saw one of those sudden momentum unwinds as investors were forced, after a weekend of reading about DeepSeek R1, to question the future. As prices fell, it forced more sellers, just as it did buyers on the way up. I traded emerging markets in the 1990s, and the door is very small in the movie theater when DeepSeek R1 yells fire.

Last week, I used the movie The Usual Suspects as an analogy for Bitcoin, and this week I will use another suspense movie that won three Oscars and is viewed as possibly the greatest of all time, Jaws. One moment, the industry was enjoying a steady trajectory, and the next, an unseen predator emerged from the depths, forcing everyone to rethink their assumptions. Just as Amity Island dismissed early shark warnings until the waters ran red, the AI world was caught off guard by DeepSeek’s impact, leading to a sudden frenzy of speculation and fear.  The unwind ignited widespread panic over the future of semiconductor demand, data center needs, and China’s position in the AI arms race. At the center of the unwind, losing close to $600 billion of market cap, was Nvidia.   

Everyone on social media, including myself, has some sort of opinion on DeepSeek. The problem is, unlike Fox vs. MSNBC, you don’t know the bias of the content creator without research. Everyone seems to have a bias when it comes to Nvidia. Either you missed out the last couple of years and have heard about it from other investors, or you have been in it and are defending it. That means, with this new news, you can still find bullish and bearish arguments if you want.

I don’t care about Nvidia the stock.  I love listening to Jensen Huang.  I do see DeepSeek itself as not surprising from a top-down perspective and something people should be thinking about now when it comes to all stocks. In fact, just before Thanksgiving, I released a Substack post titled The Moat Erosion Engine: AI’s Collision with Value Investing and the Mag 7. In that post, I wrote:

“AI introduces a new, unparalleled layer of disruption. It compresses innovation cycles and erodes long-standing competitive advantages at an unprecedented pace. My own journey, from taking Python courses five months ago to being able to speak into my phone to build websites without any coding, highlights the pace of change. Technologies and strategies that once provided resilience, such as economies of scale and brand loyalty, are increasingly vulnerable to obsolescence. AI-enabled competitors are leveraging superior analytics, automation, and adaptability to outperform traditional players, rendering even previously "untouchable" companies susceptible to disruption. While some resilient businesses may survive by innovating, diversifying, or acquiring new capabilities, these strategies are no longer guarantees of longevity in an era defined by exponential change.”

If any company fit the description within the AI industry as having a moat around it, Nvidia was that name. Clearly, based on the broad-based unwind across names, this was thought to have a moat around it.  Outside of what I wrote, this upheaval, however, should not come as a surprise to anyone listening to the smartest people in AI . Many of them have predicted that the next ten years will produce as many technological breakthroughs as the last hundred. AI development is following an exponential trajectory, where new innovations arrive with increasing speed and efficiency. The DeepSeek R1 moment is merely the latest, and certainly not the last, instance of an unforeseen leap forward redefining the rules of the game. Those who cling to outdated assumptions about gradual, linear progression are bound to be caught off guard repeatedly.

The challenge is that human cognition is inherently linear, making it difficult to process the reality of exponential change. Compounding this issue is the nature of modern media, which drowns us in a firehose of conflicting narratives. In the days following DeepSeek’s release, the news cycle was flooded with both exuberant predictions of AI democratization and dire warnings of an impending industry collapse. This cognitive whiplash further distorts our ability to objectively assess where the true opportunities and risks lie.

History provides a roadmap for how to interpret these moments. Joseph Schumpeter’s concept of creative destruction has long explained how innovation disrupts the status quo. Just as Amazon fundamentally reshaped retail and cloud computing, and as GLP-1 drugs continue to upend the consumer staples sector, the AI revolution will inevitably render entire business models obsolete while giving rise to new industry titans. Consider the rapid response to the COVID-19 pandemic, where vaccines were developed and distributed in record time—such accelerated transformation will become the norm, not the exception. The problem is, this is accelerating with AI and will only get worse.

I continue to say Bitcoin is the purest AI trade, and by now, I get pushback from almost everyone. They just don’t get the connection, so I want to use one paragraph in this paper about no moats to let people know there are moats that survive the test of time. Trust in stories like religion or gold have moats. They have survived longer than any innovation. Innovations, like currencies and governments, come and go—and now seemingly faster than ever.  Innovation has no scarcity.  Bitcoin has scarcity and is the code that allowed for digital money in the digital economy. It has outperformed every asset and is sitting at all-time highs. Solana, Ethereum, Ripple, and anything else that crypto lovers throw my way is just the same game to me. In the end, of course, I could be wrong, and maybe eventually—hopefully after I die—it will fail, but I do a video each week showing countries, states, and companies now adopting Bitcoin on top of the hundreds of millions of people in poor countries adopting it. Like soccer, Bitcoin is the World Cup of Crypto. It may be a bad technology, but it is the story people have embraced, and those are the only moats I can find over time.  It is not a crowded story based on the pushback and everyone looking for a fiat type way of valuing it, think about exponential innovation and what it truly means with scarcity. 

Sorry, I can’t write a post without mentioning Bitcoin. I do have an opinion on how these unwound trades will evolve from here for those interested.   If the value of a company and the P/E it trades at reflects the discounted cash flow over time, then P/E ratios for these companies should now, going forward, reflect the possibility that in maybe three years we will have more efficiency breakthroughs. This means, despite how their earnings go the rest of this year, I think there multiples should be lower.  It is not that these companies were likely mis-priced for the next three years, but maybe they were mis-priced for years 4-10 in the world of abundance that I have written about or purely on the disruptive nature of AI speed. If so, watch out Mag 7, because, as I keep saying, you are next. DeepSeek is unleashing a near zero-cost open-source ability for seven billion people to compete, and it doesn’t take much for a revolution of seven billion people to figure out what Jeff Bezos said: “Your margin is my opportunity.”  Seven billion overthrowing Jeff Bezos and Lauren Sanchez one dollar at a time seems like a revolution to me.

The next decade promises a convergence of more groundbreaking, disruptive technologies: robotics, AI agents, longevity breakthroughs doubling lifespan, nanotechnology, autonomous vehicles, and flying cars. DeepSeek has disrupted parts of the AI infrastructure story, but investors are already licking their losses and looking for the next momentum train. I would think these will create even higher levels of hype as we make it through 2025. For those looking to be bearish on stocks, DeepSeek absolutely speeds up the ability for companies to increase profit margins, so I just don’t see being bearish on the market as the best answer.

To survive the volatility of this new reality, one must approach it as a suspense thriller—expect the unexpected. Just as Jaws turned a peaceful beach town into a place of panic and uncertainty, AI breakthroughs like DeepSeek will continue to emerge, shocking the market and rewriting industry narratives overnight. Also, just like Jaws, there will be more sequels than people want to see. Those who recognize the pattern, anticipate the inevitable disruptions, and maintain a steady hand amid the chaos will be best positioned to capitalize on the seismic shifts of the AI revolution. I wonder if Jensen Huang is channeling his best Chief Brody and telling the Nvidia crew, “You’re gonna need a bigger MOAT.”