DeepSeek’s Disruption: The Cost Revolution Reshaping AI and Business

Causing a major ripple in BigTech is no easy feat as to challenge the existing status quo requires the development of a product or service that either improves the lives of the populace – quality; or reduces the bargaining power of the market – cost; or by just being the first – innovation. While DeepSeek may not have achieved the? last one by being the first to come up with an innovative reasoning model as this was a terrain conquered by ChatGPT, they did cause a disruption within the AI market with the release of their latest model which arguably has better performance and notably was built with a total cost less than what was used for ChatGPT.?

The significance of this disruption has not just affected the tech industry but has far reaching effects into the political terrain. It has sparked debates, which has not been seen since the nuclear arms race and this debate currently lies between the two super-powers – China and the USA with government involvement in the respective private companies that are responsible for building these products – DeepSeek and Open AI.

However, the questions on every business owner’s lips at this time is ‘what does the disruption mean for businesses going forward’, and with the alleged reduced cost of building ‘will it be possible for businesses to build their own AI models in the future’.?

What does DeepSeek’s disruption mean for businesses going forward?

This is not the first time an AI model would be released after OpenAI’s ChatGPT as we’ve had Google’s Gemini and Anthropic’s Claude. So why is the release of DeepSeek different from the others? This may be because it did the following, days into entering the market officially:

• It was said to have a total cost of $6 million as compared with the development ChatGPT 4.0 which cost $100 million.
• These cost dynamics affected Nvidia’s stock price which reduced by a whopping $600 billion the Monday after its release.

It has been highlighted that the most impactful of these is the ‘cost of building’ which affected the stock price of Nvidia as there is now the counter-belief against the original belief that to build AI models, it requires expensive computing resources, and complex microchips produced by the company. The DeepSeek team has broken the mold with their creation as they were able to allegedly achieve better performance without the need for the same high level and costly resources.?

It must be noted that this reduction in building cost has had a cascading effect on the subscription prices quoted for businesses who want to subscribe to the product as a means of integrating it into their processes either for emails or for more complex tasks such as coding as compared to ChatGPT. Currently, the subscription for DeepSeek is free while ChatGPT has a free version with limited features and a premium version priced at $20 per month. For access to API, DeepSeek charges? $0.48 per million tokens while ChatGPT charges $3–$15 per million tokens, depending on model.? However, it has been argued that this current DeepSeek’s costing system may not survive the long term especially as the number of users increase in order to maintain quality.?

Will it be possible for businesses to build their own AI models in the future?

Another perspective brought to light with this whole DeepSeek costing? is the possibility of democratizing the ability of businesses to build their own AI model in-house as cost is no longer an issue. This is an especially promising outlook for businesses as they not only get to experiment and fully integrate AI and all of its features directly into their processes instead of the current secondary state. Also, it helps to safeguard? data privacy which is still a shoddy area for the current AI’s in the market as there are alleged misuse or use without consent of data to train their models. This can be detrimental for businesses working with highly confidential data opening them up to liabilities.?

This outlook seems to be favored by pundits within the industry and even beyond despite? the fact that on the other hand there is still a lot of speculation and uncertainty surrounding not just the building of AI’s but also what they can do.

So what should you do??

With so much uncertainty still surrounding the release of AI models, an option as a business owner is to do ‘nothing’ and wait till the dust settles. However, you run the risk of being left behind by your competitors who are using these AI models either to improve their effectiveness or their speed in delivery of their product or services to customers.?

A more effective approach is to do an appraisal of your businesses needs and see which one of these leading AI models best fit into your business process and produces the most value. It is important to be holistic in your approach and not just focus on the cost but also their performance. At the end of the day, to get the most value out of AI models, it may require a combination of two or more models to achieve a particular goal.

Overall, the outlook for reasoning models is great and the healthy competition is likely to make the industry accessible for more companies.