6 min read
DeepSeek and Destroy
How China’s AI upstart is disrupting the market
Artificial Intelligence (AI) has become more than just a technological race – it’s now a matter of economic leverage and geopolitical strategy. Enter DeepSeek, a China-based AI lab backed by the hedge fund High-Flyer, which has burst onto the scene with a model that’s making waves for its combination of low cost, strategic marketing and open-source rhetoric. Discover how DeepSeek, a China-based AI lab, is disrupting the AI market with low-cost models, strategic marketing, and open-source rhetoric
But, before we get into it, a reality check.
Review
Beware
Watch
With its R1 reasoning model priced 27 times cheaper than OpenAI’s O1 API, DeepSeek is attempting to fundamentally shift the economics of AI. The company’s approach mirrors successful Chinese tech disruptors like Temu, BYD, and TikTok, who have historically undercut Western competitors with ultra-low pricing and a community-first approach.
Yet beneath this disruptive energy lie some critical questions:
- Is DeepSeek genuinely open-source?
- What does its rise mean for U.S. AI companies?
- And, most importantly, what happens when a low-cost AI model potentially sends user data to China?
The DeepSeek playbook: Cheap, fast, and unavoidable
DeepSeek’s strategy is straightforward: offer AI at a fraction of the cost of U.S. competitors, win over developers, and create long-term market dependency before profits even matter.
This is the same tactic that made TikTok a social media juggernaut and turned Chinese EV brands like BYD into global threats to Tesla.
The formula is simple but effective:
Flood
Gain
Force
Time
DeepSeek is applying this method with precision, and it’s working. Developers, startups, and budget-conscious companies are eager to reduce AI costs, and DeepSeek’s offer is hard to ignore. The impact is already visible – DeepSeek is at the top of the iOS app store, signalling strong early adoption. But this pricing strategy raises a fundamental concern: Is it sustainable?
Operating at a loss to gain market share isn’t a new strategy, but it’s particularly risky in AI, where training and maintaining models require massive computational power. OpenAI and Anthropic cannot simply slash prices without gutting their revenue models, meaning DeepSeek is forcing them into an uncomfortable dilemma: play DeepSeek’s game or risk losing ground.
When the chips are down…
With the U.S. restrictions on chip sales to China since 2021, “Chinese AI developers have shared their work with each other and experimented with new approaches to the technology”. Whether DeepSeek’s founder stockpiled chips or not, the fact that another global chip shortage could be looming as soon as 2026, underscores the importance of strategic resource management in AI development. With China currently being the only market that pursues LLM efficiency owing to chip constraint this could impact how AI innovation and competition evolve globally.
The ability to develop efficient models with limited resources may change country priorities, outlooks and may end up being a critical advantage, potentially reshaping the AI landscape and challenging the dominance of established players.
The ‘Open-source’ illusion
DeepSeek has positioned itself as an open-source alternative to proprietary models like OpenAI’s GPT series. This has won them favour with the developer community, which has long been frustrated by AI companies locking models behind APIs and strict licencing terms.
But the reality is more complicated.
-
- DeepSeek has released model weights, but not training data or training code – a move that mirrors Meta’s LLaMA strategy rather than true open-source principles
- This means that while companies and developers can use DeepSeek’s models for free, they cannot independently verify how the models were trained or what biases exist within them
- By withholding these critical components, DeepSeek maintains control over its technology while benefitting from the “open-source” goodwill
This approach is clever as it allows DeepSeek to gain traction as a champion of open AI while still holding onto its proprietary advantages. But it also means developers should be wary of trusting too much in the promise of openness when the core components remain out of reach.
Winning over developers: The ‘Hero’ narrative
Beyond pricing and open-source positioning, DeepSeek is tapping into something more powerful: developer loyalty.
The company is framing itself as the scrappy underdog, standing up against AI giants like OpenAI, Anthropic, and Google. This kind of “us vs. them” messaging has worked before – TikTok gained traction by presenting itself as the rebellious alternative to Facebook and YouTube, and Temu is doing the same against Amazon.
DeepSeek is playing a similar game, making developers feel like they are part of a movement – one that challenges the monopolization of AI by Silicon Valley. The result? A growing community that is seemingly eager to support DeepSeek, even if its long-term intentions remain unclear.
This puts OpenAI and Anthropic in a difficult spot. They now face pressure to either lower costs (which isn’t easy for high-end AI models) or innovate faster (which comes with its own risks). DeepSeek is controlling the narrative, and U.S. firms are now reacting rather than leading.
Shares in AI-related companies based in the US, such as Nvidia, Microsoft and Meta were down on Monday morning – and the development knocked European share prices.
ASML, the Dutch chip equipment maker, saw its share price tumble by more than 10% while shares in Siemens Energy, which makes hardware related to AI, plunged by 21%.
The security question
Of course, this wouldn’t be a discussion about Chinese AI without addressing the elephant in the room: data security.
DeepSeek’s hosted API version processes user queries, meaning any text inputted into its system could potentially be accessed or stored on servers tied to China.
This raises alarms for companies handling sensitive business information, government agencies, and developers concerned about data privacy.
The U.S. has already scrutinised TikTok for similar concerns – DeepSeek is likely next.
DeepSeek’s ability to collect massive amounts of Western training data via consumer and enterprise applications could also reduce China’s dependence on web scraping and unauthorised data collection. Instead of struggling with firewalled Western content, DeepSeek could simply absorb the data directly through user interactions.
This possibility won’t go unnoticed by regulators or lawmakers. If DeepSeek gains serious traction, expect investigations, restrictions, or even outright bans, particularly for industries handling sensitive information.
But let’s see the devil is in the detail and as you can imagine a Chinese model will be controversial for many uses. Still it is a cold shower and a dose of reality for a sector that probably needed it.
The future: A collision course with U.S. AI firms
So, where does all of this lead?
DeepSeek’s expansion will inevitably force U.S. AI firms to react. OpenAI, Anthropic, and Meta cannot afford to ignore its rapid rise, but their options are limited. Lowering prices too much could undermine their own revenue streams, while focusing on high-end, premium AI models might alienate cost-sensitive users. Rather than engaging in a direct price war, they will likely double down on differentiation, emphasising higher-quality models, better reasoning, and greater reliability.
Government scrutiny is all but inevitable. The U.S. has already imposed restrictions on Chinese tech firms over data security concerns, and DeepSeek could soon face similar regulatory pressure. If its adoption grows significantly, it may encounter the same obstacles that TikTok did. At a minimum, government agencies and major corporations will hesitate before entrusting DeepSeek’s AI with sensitive operations.
Adoption patterns will also be divided. Startups and independent developers will likely flock to DeepSeek, enticed by its affordability. Larger enterprises – especially those in finance, healthcare, and government – will be far more cautious, wary of the potential risks. The real question isn’t whether DeepSeek will gain users – it’s whether it will ever be fully trusted by the broader AI ecosystem.
The bottom line
DeepSeek isn’t just another AI startup – it’s a calculated disruption.
With its ultra-low pricing, strategic use of open-source marketing, and developer-first approach, it has already put pressure on the biggest names in AI. But beneath this appealing façade are serious concerns about data privacy, transparency, and long-term control.
Yet DeepSeek is just one piece of a much larger puzzle. China’s AI push is accelerating, with companies like Moonshot AI and its Kimi matching industry benchmarks. The race is no longer just about innovation – it’s about control, influence, and long-term dominance in a technology that will shape the future of economies, industries, and national security.
For developers and startups, DeepSeek is a game-changer – affordable, powerful, and hard to ignore. But for those focused on security and sustainability, it’s a wild card – one that could level the playing field or reshape AI in unpredictable ways.
