Chinese AI startup DeepSeek has made waves after releasing a cost-efficient chatbot to rival the likes of ChatGPT, topping iPhone app stores in numerous countries. News of the tech has spread doubts over the valuations of Western AI companies, and the share prices of major tech firms have experienced some of the steepest stock price drops in US market history. To gauge DeepSeek’s impact, FutureWeek asked seven industry leaders across marketing, AI and investment, to weigh in on what it means for the future of AI and marketing.
Wesley ter Haar, Co-founder, Monks
“I’ll stay away from the geopolitical implications, and keep it practical. DeepSeek confirms that there is no wall, that there is no moat and that the line truly just keeps going up and to the right (jumping over the ‘trough of disillusionment’ of a traditional hype cycle). I’d recommend everyone still skeptical to spend some time with DeepSeek’s chain of thoughts to truly understand the trajectory we’re on. Abundant (and cheap) intelligence, ever more generally applicable, fundamentally changes our industry and many more. Now we wait for OpenAI o3.”
Jingjing Xu, Managing Director, Fuel Ventures Asia
“DeepSeek’s success highlights the power of collaboration in resource-constrained environments, such as China’s limited access to advanced US chip technology. European startups, already experienced in navigating similar constraints, could adopt this approach, leveraging DeepSeek’s breakthroughs to explore new applications and business models. This could foster partnerships within Europe, creating a more resilient and competitive AI ecosystem.
Additionally, the buzz surrounding DeepSeek may invigorate European investors, encouraging greater financial support for AI startups. However, this surge in Chinese AI innovation also intensifies pressure on European startups to innovate cost-efficiently, as they must now compete in a market increasingly defined by affordability and scalability. Balancing these opportunities and challenges will be critical for European players aiming to thrive in this rapidly evolving landscape.”
Dom Couldwell, Head of Field Engineering EMEA, DataStax
“This is the first big step forward making interference and training cheaper and not just bigger, faster and stronger. This came out of China which had been assumed, in some parts, to be behind the curve. That’s clearly not the case. Any reduction in the cost of AI is obviously a good thing for enterprises but this also reinforces the need to consider LLMs as a commodity and be able to switch to different models as they emerge. Enterprise spend on AI won’t necessarily reduce because of the change but they will be able to do more and may need to consider the balance between using off the shelf versus self trained models as model training costs reduce.”
“Will there be a surge in demand for more commoditised hardware to support AI versus specialised solutions? Yes and no. For simple models they may be able to run on less specialised hardware but we’ve not gotten to the limit of innovation on the software side. Many solutions will still be hardware limited for now and need more specialised hardware.”
Rob Webster, Co-founder, TAU Marketing Solutions
“DeepSeek has certainly set the cat amongst the pigeons in the AI space. Their new open-source model, DeepSeek-R1 is a breakthrough in making advanced AI much cheaper and more accessible. While it doesn’t reinvent what AI can do just yet, it’s a big step towards democratising capabilities and proof that the future isn’t all about the established US giants – and they won’t have it all their own way.
“Not only does this lower costs, but also energy consumption – a boon for sustainability concerns. Its open-source approach could also benefit companies who want to host their own models for security. However, as with TikTok, the Chinese ownership does raise questions about potential influence. For marketers, this doesn’t change the art of the possible, yet it does lower barriers to adoption. Its also a huge signal as to how quickly the space can change and you can see the established giants like Open AI, Google and Meta looking to catch up.”
Ines Clark, Founder and CEO, MartyAI
“This reckoning highlights the undeniable advantage that nimble startups have over larger tech giants. With its homegrown talent, American and European tech can reclaim their leadership, especially given the challenges Chinese-born tech faces when expanding globally. However, this also underscores the need to invest more in the early-stage startup ecosystem to foster broader innovation and growth.”
“Looking at the impact on companies’ strategies, this creates a need to implement an open-innovation approach to engage quickly with early-stage startups and differentiate from the competition. Deprioritising this for seemingly more pressing matters creates long-term material risk.”
Ben Wood, Performance Director, Hallam
“DeepSeek’s R1 launch is undeniably a pivotal moment in AI development, particularly the low-cost involved in training the model. For marketers and agencies, it’s a signal that the AI tools and models we rely on are about to become even more powerful and accessible. Advancements in AI may no longer be accessible only to large network agencies, and we could instead see access to powerful AI models democratised for agencies of all sizes. For independent marketing agencies, this should open up opportunities to harness more affordable AI tools.
“With new players like DeepSeek pushing the boundaries, and giants like OpenAI and Meta already responding, we can anticipate a surge in accessible AI-driven marketing solutions in the near future, heightening client expectations around innovation, speed of execution, and value.”
Ankuj Arora, CEO, Morphiq
“First of all, a firm tip-of-the-hat where its due. Kudos to the DeepSeek team for a remarkable engineering feat, and for dispelling the notion that compute capital is a fundamental need for building top class foundational models.
“Hard to see how this will change the ‘Madtech’ landscape, as writing scalable software is still a few notches harder than writing code (RBS’s Bo experiment comes to mind). Having said that, thanks to genAI powered code builders, the cost of prototyping, iteration and therefore startup capital is meagre.
“With best-in-class inference merely a local deployment away, it feels like traditional SaaS architectures might be looking at a major overhaul. In some ways it feels like the Facebook days all over again, where using off-the-shelf technologies, one college kid coded the F in FAANG from a dorm room in 40 days. So the question remains: Who’s building the Facebook for the new world order?”