What Does DeepSeek Mean For Nearshore Operators?

DeepSeek’s emergence in the tech world and stock markets is accelerating the race for dominance in large language models (LLMs). AI development is now becoming much like the space race of the past. This time, however, the competition is between American and Chinese companies. The impact on global productivity and the speed of technological transformation is expected to be immediate and significant.

But what does this mean for nearshoring service providers? In practical terms, DeepSeek’s introduction does not fundamentally change AI adoption—it is simply another AI model. The key distinction, however, is that it is owned by High-Flyer, a Chinese hedge fund.

Argentine tech expert and professor Carlos Pallotti sees a direct impact on the business process outsourcing (BPO) sector. He notes, “This AI development is definitely influencing the BPO world.”

The main driver of this impact is competition. “Advancements like these push companies to optimize their processes. With each new breakthrough, reengineering workflows often becomes necessary,” Pallotti explains.

Yet, the most profound effect of DeepSeek on the industry lies in its financial implications. Equipped with an efficient database, DeepSeek AI enhances productivity like other AI models, but its real advantage is cost. The price per token of DeepSeek-R1 is less than one-thirtieth of what OpenAI charges for GPT-4-turbo.

Palotti commented: “Tech developments like DeepSeek can reduce costs significantly. When they do, new service areas can enter the world of massive AI adoption.”

As always, language barriers are key: Worldwide, IOS App Store’s most downloaded ranking is now led by DeepSeek.

However, this doesn’t change the fact the system isn’t as effective for everyone.

According to Palotti, “All AI models have biases. Before worrying about data privacy, one should be cautious and beware of possible skews in final results of their usage.”

An API, or Application Programming Interface, serves as the front-end of any large language model (LLM)—the interface where users interact with the AI by entering prompts. According to analyses by GitHub and DataCamp, the APIs of the Chinese-developed DeepSeek model are currently optimized only for Chinese and English.

This underscores the growing importance of coding proficiency and language skills in either Chinese or English for those looking to leverage emerging AI technologies. For nearshoring companies, this also means DeepSeek should be used exclusively in these languages to minimize the risk of AI-driven reasoning errors.

The fine print: There are two common ways to use public large language models: desktop versions or apps and browser APIs. Both groups are distinct in three aspects: whether they require internet connection for all operations (which desktop versions don’t), their processing power, and their privacy agreements.

DeepSeek’s privacy policy states:

“The personal information we collect from you may be stored on a server located outside of the country where you live. We store the information we collect in secure servers located in the People’s Republic of China.”

In the case of online apps and browser API, that data recollection never stops.

Stoyan Mitov is CEO at Bulgaria-based Dreamix, a B2B software development company with global operation. He advises BPOs and tech developers to review data privacy agreements and active legislation. “Laws like GDPR and CCPA are in place to protect consumer privacy and ensure companies handle data responsibly. Ignoring these regulations can lead to serious consequences—fines, reputational damage, and loss of trust from your customers. By ensuring that your AI tools meet these standards, you protect both your company and your customers while also increasing trust in your business.”

According to Nearshore Cyber’s CEO, Peter Schawacker, businesses should be cautious when feeding data to DeepSeek:

“Any company that deals with sensitive data should run a local (desktop) version of the LLM, regardless of its origin. The SaaS generative AI providers cannot guarantee that prompts or data won’t be used to train their models. Best way to be sure you have a relatively secure AI system is to run a version on infrastructure that you control and to restrict traffic between the LLM and the Internet.”

This past week, U.S.-based Perplexity and Microsoft’s Azure made strategic moves in response to DeepSeek’s growing presence. Leveraging both the excitement surrounding the model and concerns over Chinese access to corporate data, both platforms announced they would integrate DeepSeek into their own open-source models through their APIs. Perplexity, in particular, is emphasizing a strong data security message: “Your data never leaves Western servers. The open-source model is hosted completely independent of China.”

Should companies using DeepSeek worry about facing a “TikTok-style” crackdown? In the short term, no. However, accessing DeepSeek’s services directly through its API presents a far greater data security risk than TikTok ever has.

Both Zack Doffman (Forbes) and Tech Marketing Expert Pam Aungst Cronin have written in-depth articles regarding DeepSeek’s security considerations in the age of the big data race. Both agree that DeepSeek and large language models generally fall outside the scope of the Protecting Americans from Foreign-Adversary Controlled Applications Act. (PAFACA, the “TikTok Ban”).

The main reason is it was designed exclusively for items that meet all the law’s definition of a popular social media application. In her article, Cronin says “the short-sightedness of the PAFACA law left the United States extremely vulnerable to national security risks that are light years above and beyond those presented by TikTok”.

Doffman’s analysis goes so far as to say “this is strategic in a way TikTok never was.” He points out how the Language Model’s privacy policy is also clear on the fact the data they may collect includes account information and the documents and inputs you feed the model with.

However, Schawacker is on the side that believes espionage is not a real danger: “[The fact you are giving up data] is as true for Meta’s Llama as it is for DeepSeek. If DeepSeek is a trojan horse, so is every US-owned social media platform. Security is not the issue, nor is China. Blaming China is an attempt to deflect criticism by whipping up panic.”

NSA’s take: DeepSeek’s announcement came shortly after Trump’s handling of the TikTok crisis and the launch of Stargate Program. With these three major tech developments unfolding in quick succession, we seem to be headed into ever more disruptive times for tech development.

However, rushing to integrate the latest AI tools in response to geopolitical shifts could be a costly mistake. The combination of sweeping geopolitical changes and record-breaking stock market fluctuations can put immense pressure on business leaders. Yet, despite the urgency to adopt cutting-edge tools, experts caution that we should take a lesson from Aesop’s fable of the hare and the tortoise—steady and strategic progress often wins the race.