The DeepSeek Disruption

Monday sent shockwaves through the US tech sector as DeepSeek, a Chinese AI startup, delivered an unsettling blow to the industry. Giants like Nvidia, Microsoft, and Google saw their valuations dip during pre-market trading as investors faced the truth: cutting-edge AI no longer demands a $17 billion budget or a Silicon Valley-sized workforce… or so we thought!

DeepSeek isn’t alone. Other Chinese players, such as Bytedance’s Doubao-1.5 and Moonshot’s Kimi k1.5, are gaining momentum, signalling a shift in the global AI race. The competition is fierce, and China’s knack for undercutting prices is setting the stage for a heated AI price war.

Investors are now questioning whether US tech stocks are overvalued. Years of soaring valuations, fuelled by AI hype, are being reevaluated as China’s cheaper, equally effective alternatives step into the spotlight. Meanwhile, the ripple effects of this shake up spilt over into crypto markets, further denting investor confidence.

But not everyone is folding. After a 20% stock tumble, Nvidia issued a pointed reminder: DeepSeek’s success only underscores the critical importance of its GPUs. “Inference requires significant numbers of Nvidia GPUs,” the company noted, reinforcing the idea that even AI revolutions still rely on their tech.

Amid the chaos, one thing stood out: retail investors saw an opportunity in the bloodbath. According to Vanda Research, they poured a record $562.18 million into Nvidia stock during the DeepSeek selloff. It’s a reminder that in every capitulation, there’s a buying opportunity.

Who is behind DeepSeek and how they cracked the AI code?

DeepSeek was founded in late 2023 by Liang Wenfeng and has quickly emerged as a top player in the AI industry. Liang, a former hedge fund manager and machine vision expert from Zhejiang University, combined his deep understanding of computational efficiency with his experience in managing large-scale operations to create a lean, agile company. With a team of around 200, DeepSeek developed an AI model, DeepSeek-R1, at a fraction of the cost and scale of its competitors. By focusing on software efficiency rather than sheer hardware power, DeepSeek achieved what many thought impossible. Leveraging innovative techniques such as advanced neural network pruning, dynamic computation pathways, and precise parameter fine-tuning, the team significantly reduced the computational requirements for their model. This allowed DeepSeek to produce performance levels comparable to leading models from companies like OpenAI, but at a development cost of only $5.6 million, far below the billions typically spent on such projects.

A key aspect of DeepSeek's strategy was its ability to balance hardware dependency with software-driven optimisations. While the company relied on tens of thousands of Nvidia's H800 and H200 GPUs, which are less advanced than the H100 clusters widely used in the U.S., it employed proprietary techniques to maximise efficiency. Custom communication schemes between chips, memory-saving techniques, and reinforcement learning algorithms enabled DeepSeek to achieve high performance with less powerful hardware, showcasing a shift toward more accessible and cost effective AI development. These innovations allowed DeepSeek to dramatically undercut its competitors in terms of API pricing.

The announcement of DeepSeek-R1 had profound implications for the industry, most notably sending Nvidia’s market value plummeting by nearly $600 billion in a single day. This historic drop reflected investor fears that DeepSeek’s model could undermine demand for Nvidia’s high-end GPUs, which are a cornerstone of the AI hardware market. However, I think this was largely an overcorrection. At the end of the day, DeepSeek still relies on Nvidia’s hardware, albeit in a more efficient manner, and Nvidia remains deeply entrenched in the broader AI space.

The geopolitical implications of DeepSeek’s success were also significant. President Trump described the company’s achievement as a wake up call for the United States, urging the country to prioritize AI innovation to remain competitive. His administration’s $500 billion Stargate Project underscores the urgency of maintaining technological leadership in the face of rising competition from China. Meanwhile, OpenAI’s CEO, Sam Altman, welcomed the new competition, noting that it would invigorate the industry and push companies to innovate further.

It is safe to say that DeepSeek’s accomplishments mark a massive moment in the AI industry. By demonstrating that cutting-edge results can be achieved without massive budgets or excessive reliance on high end hardware, the company has set a new standard for efficiency and accessibility. While Nvidia and other industry giants may face short-term challenges, the long-term impact is likely to be a more dynamic, competitive AI landscape driven by innovation and cost-effective solutions, which is incredibly bullish!

So what does this mean for the Mag 7 and broader emerging Tech?

DeepSeek's arrival has brought both challenges and opportunities for the MAG 7. As we saw yesterday, the initial impact was a sell-off in tech stocks, with Nvidia bearing the brunt of the market's reaction (almost $600 billion in Market Cap wiped out). Nvidia's drop in value highlights investor concerns over the reduced demand for its high-end GPUs, as DeepSeek's cost efficient AI models demonstrate that cutting edge performance can be achieved with less powerful hardware. This shift threatens the traditional business models of companies that rely on expensive, hardware-intensive solutions, forcing them to reconsider their strategies. Additionally, scepticism about DeepSeek's open-sourced model raises concerns over intellectual property (IP) theft and potential misuse, particularly given geopolitical tensions between the U.S. and China. Furthermore, DeepSeek's storage of data within China has intensified fears about security and privacy, creating hesitancy among U.S. entities regarding foreign-developed AI systems.

Despite these challenges, I believe there are incredibly positive implications for U.S. tech and the broader AI race. Increased competition is pushing giants like Google, Microsoft, and Meta to accelerate their innovation, fostering a more dynamic and competitive environment. DeepSeek's success in delivering high performing AI models at a fraction of the cost is a wake-up call for U.S. companies to optimise their operations and explore more efficient development methodologies. This cost efficiency will also likely lower barriers for smaller players to enter the market, democratising access to advanced AI technologies. Ultimately, these innovations are likely to drive a new wave of breakthroughs in AI efficiency, accessibility, and fairness… something all emerging sectors need! Without competition, sectors lack robustness.

Beyond the MAG 7, DeepSeek’s influence extends into emerging areas such as decentralised AI (something we have discussed at length in previous newsletters) and blockchain-based AI platforms like Bittensor (Bittensor’s decentralised AI network promotes open-source solutions, offering an alternative to the limitations of centralised AI).

Decentralised AI, which thrives on principles of transparency, collaboration, and accessibility, aligns closely with DeepSeek’s open-source ethos. By open-sourcing their model, DeepSeek empowers global communities to contribute to AI development, mirroring the community driven nature of platforms like Bittensor. This approach reduces centralisation in AI innovation, promoting a more diverse and resilient ecosystem where developers worldwide can collaborate on equal footing.

For Bittensor specifically, DeepSeek’s advancements serve as a proof of concept for cost effective, decentralised AI development. The success of DeepSeek validates Bittensor's mission to democratise AI. It also highlights the growing viability of decentralised systems that prioritise efficiency and community engagement.

Moreover, blockchain’s inherent advantages in transparency and security address key concerns raised by centralised AI models. Blockchain platforms offer auditable, decentralised data storage and processing, ensuring data integrity and reducing risks associated with centralised control. As the AI landscape becomes more globally collaborative, blockchain projects like Bittensor facilitate cross-border development, breaking down geographical and political barriers to innovation. The whole narrative is to push the idea of open source; this is what AI should be!

DeepSeek’s impact signifies a broader trend toward true open-source, efficient, and inclusive AI development. Introducing competition into the space reduces the dominance of a few tech giants and creates space for emerging players and technologies to thrive. so while we had the doom and gloom of yesterday’s market turmoil we need to remember that DeepSeek’s training model is open-source which means the MAG7 can use it to scale and adapt and at the end of the day the Nvidia chips still need to be used… All in all, this is bullish AI but most notably this is bullish on the cross-section of blockchain and AI technologies.

So what am I doing?

Overall, I'm feeling bullish about the future of AI markets. The rise of open-source AI, especially with China’s DeepSeek, means U.S. companies will need to become more efficient to stay competitive. These developments also bring powerful new AI applications to the table, which is exciting. Sure, the market might have overreacted with concerns about a potential AI bubble burst, as it tends to do, but the shift in AI power could lead to significant growth for open-source AI, as long as innovation isn’t regulated out of existence.

Short-Term

Looking at the short term, I’m paying close attention to the data coming in this week. The market often reacts in a certain way leading up to the FOMC meeting:

• Monday tends to be a pre-FOMC dip (As we saw, accelerated by the DeepSeek Catalyst).

• Tuesday is a consolidation day, as traders position themselves ahead of the big announcements.

• Wednesday is the FOMC meeting, and if everything aligns as expected, it could solidify the local bottom.

The key takeaway here is that there’s strong sentiment indicating no rate cuts from this FOMC meeting, although there’s some optimism following Christopher J. Waller’s recent remarks about potential rate cuts in March. The market is recalibrating in response to this, adjusting expectations accordingly. What’s also interesting is that the additional impact of DeepSeek has led to further shifts in investor positioning, especially as the end of the month approaches and positions are being rounded out. This creates a double catalyst for volatility and liquidity.

Waller expressed optimism about the disinflationary trend, suggesting inflation could return to 2% more quickly than expected. With the upcoming FOMC meeting, GDP data, and the PCE price index this week, all eyes will be on these events. A dovish tone from the Fed, especially paired with favourable GDP and inflation data, could trigger a relief rally and a broader bullish continuation for risk!

Year End Outlook

Looking further ahead at general risk markets, I’m still bullish for 2025, especially considering the fundamental shifts happening right now. The introduction of DeepSeek is just one example of how AI and tech are evolving, and I believe this is only the beginning.

The intersection of blockchain and AI is creating an exciting rotation into crypto markets. This convergence presents incredible opportunities, but it’s important to recognise that this bull market is unlike any we’ve seen before. The sheer explosion of tokens in recent years highlights why fundamentals will matter more than ever. Consider this: in 2013, there were only 500 tokens… fast forward to 2021, and we had around 3 million tokens. Today, the projections suggest that by the end of the year, we’ll hit around 100 million. This growth has fundamentally reshaped the market landscape.

With so many tokens now competing for attention, liquidity is being diluted, spreading capital thin across millions of projects. As a result, we’re seeing fewer large-scale pumps and weaker long-term growth for less robust projects. The rise of meme tokens has added another layer of complexity, as new launches occur daily, often with little utility, leading to quick pump-and-dump cycles and fragmenting investor focus. Meanwhile, the market itself has matured. Investors are becoming more discerning, shifting their focus to utility and solid fundamentals, and there’s less tolerance for hype-driven projects without clear value.

To thrive in today’s "Wild West", discipline and a long-term perspective are essential. I believe investors should focus on projects with strong fundamentals and real-world utility, particularly at the intersection of AI and blockchain, where solutions like decentralised computation, secure data sharing and distributed GPUs are gaining traction. With global trends aligning, analysts are largely forecasting double-digit returns for the S&P 500 this year, suggesting an incredibly strong year for risk markets overall (crypto typically correlated to traditional risk).

In addition to the bullish stance on the year, it is also important to note the technical benefits that came out of yesterday’s market sell-off, namely the increase in market breadth... Currently, 64% of S&P 500 companies are trading above their 200-day moving average, reflecting a drastic increase from the early Jan low of 49%. Additionally, 60% of firms are now above their 50-day moving average, up significantly from around 17% three weeks ago. So even with the S&P 500's drop yesterday, driven by declines in big tech, the majority of stocks closed higher, signalling a healthier and more resilient market foundation beneath the surface.

It’s a volatile time, but I’m focused on the bigger picture: the long-term growth of risk assets, particularly in AI and tech. All eyes are on the FOMC this week to see how they guide us, but I remain confident that 2025 holds a lot of potential.

Key Takeaways:

1. DeepSeek disrupts AI with cost-efficient models, challenging U.S. tech giants and sparking a global AI price war.

2. Nvidia remains essential despite a $600 billion market cap drop, as retail investors bet on its long-term role in AI.

3. DeepSeek’s success highlights the rise of software-driven AI innovation, reducing reliance on high-cost hardware.

4. Increased competition drives AI and blockchain innovation, fostering decentralisation, efficiency, and accessibility.

5. Market resilience improves as S&P 500 breadth strengthens, signalling healthier foundations amid tech sector volatility.

Don’t miss out - join us for free!

Get exclusive access to top-tier trading insights and strategies from the institutional front lines. Each week, we deliver actionable market trends, cutting-edge risk management techniques, and lessons learned from both the retail and institutional worlds.

Subscribe now to Peak Performance for your backstage pass to the elite market mindset!