Startup DeepSeek launched their AI-chatbot DeepSeek-R1 in January 2025 sending shockwaves throughout the tech world and the stock market.
The DeepSeek chatbot is open-source unlike its popular counterpart ChatGPT, which is closed-source. Open-source models have publicly available weights which can be viewed and edited by anyone. This allows companies, especially startups, to readily take these models and then develop and fine tune them further, which can speed up innovation. Additionally open-source models can be cheaper to build and deploy. For example, the training cost of DeepSeek is reported to be significantly lower than many of its competitor large language models.
Other chatbots have been presented as open-source previously, such as Meta’s Llama 3.2, but arguably none have generated the same level of initial interest as DeepSeek. However, with such rapid development in the industry, this may soon change. There has already been a flurry of activity following the DeepSeek release. For example, a team from University of California Berkeley quickly developed a small-scale reproduction of DeepSeek R1-Zero for around 30 dollars coining their project “TinyZero” and showing that creation of AI models need not be expensive or take many years.
Alongside this, we are seeing increased investment in the AI sector more broadly. For example, hot on the heels of the launch of DeepSeek, the US announced the Stargate Project which will see a 500 billion dollar investment to build AI infrastructure in the US. The hope is that “this infrastructure will secure American leadership in AI, create hundreds of thousands of American jobs, and generate massive economic benefit for the entire world”. The project boasts support from well-known companies such as Microsoft, SoftBank, Oracle, OpenAI, and NVIDIA. However, historically, periods of significant investment have not only benefitted the established players but have also paved the way for the success of newcomers.
The DeepSeek release and the response to it has certainly shown that emerging players can generate huge amounts of interest and discussion, and that the AI industry is susceptible to radical change. We expect to see further disruption over the next few years, with a range of new technologies being developed.
As ever, where there is significant competition and investment, there is often a focus on protecting intellectual property. It is therefore likely that any boom in AI development will be accompanied by a corresponding boom in AI patent filings. Moreover, while there has been mixed success historically with protecting innovations in core AI, patents are particularly suitable for protecting specific applications of AI. As a result, we envisage increasing patent filings from not only the AI developers themselves, but also those building upon the models for downstream applications.
Get in touch with our AI team today.
