What is DeepSeek? The Chinese AI app challenging global tech giants

DeepSeek: Everything you need to know about the AI chatbot shaking up the industry

China’s viral AI app is redefining innovation, sparking U.S. bans and Wall Street disruption

DeepSeek, a Chinese AI startup, has surged to international attention after its chatbot app topped the Apple App Store and Google Play charts — and rattled the nerves of U.S. policymakers and tech rivals in the process.

Backed by High-Flyer Capital Management, a quantitative hedge fund with roots in AI-driven trading, DeepSeek’s rise has been fueled by cutting-edge models, an aggressive talent strategy, and a product philosophy that prioritizes efficiency and scale over traditional business economics.

From trading algorithms to viral AI

DeepSeek began as a research lab inside High-Flyer Capital in 2023 and later spun off into its own company. Co-founded by AI enthusiast Liang Wenfeng, the lab quickly distinguished itself by building proprietary data center clusters and focusing on compute-efficient model development — despite facing U.S. export restrictions on high-end chips.

To circumvent the lack of access to Nvidia’s powerful H100 chips, DeepSeek trained its models using the less-capable H800, tailored for the Chinese market. Nevertheless, it has delivered breakthrough performance across multiple benchmarks.

The DeepSeek model lineup

The company's first wave of models — DeepSeek LLM, DeepSeek Coder, and DeepSeek Chat — launched in late 2023. But it was the DeepSeek-V2 and V3 releases in 2024 that catapulted the firm into the international spotlight. These models are not only powerful and commercially viable, but they also undercut rivals in cost, prompting competitors like Alibaba and ByteDance to drop their prices.

DeepSeek V3 outperforms many popular models, including Meta’s LLaMA and even rivals like OpenAI’s GPT-4o, according to internal benchmarks.

Then came DeepSeek R1, the company’s flagship reasoning model, launched in January. Designed to fact-check itself and reason through multi-step logic, R1 is more accurate in science, physics, and math — albeit slightly slower than conventional models.

In May, DeepSeek released the R1 update to Hugging Face, where over 500 derivative models have already been created by the developer community.

Efficiency over profit

DeepSeek’s approach is disruptive by design. It offers models under commercially permissive licenses, and prices services below market value. The company claims that efficiency breakthroughs allow it to compete without VC funding — a strategy that has drawn both admiration and skepticism.

According to Hugging Face CEO Clem Delangue, DeepSeek's models have collectively been downloaded over 2.5 million times.

However, the models remain subject to Chinese content regulations, meaning they will refuse to answer questions about politically sensitive topics such as Tiananmen Square or Taiwan — a point critics say undermines their neutrality.

Political backlash and global bans

Despite DeepSeek’s technical excellence, its growing influence has sparked geopolitical concerns. In March, the U.S. Department of Commerce instructed staff to ban the app from government devices. New York State and South Korea followed with similar bans.

OpenAI has accused DeepSeek of being “state-subsidized and state-controlled,” while Microsoft’s Brad Smith confirmed that employees are prohibited from using DeepSeek due to concerns over data security and propaganda.

Meanwhile, Nvidia CEO Jensen Huang described DeepSeek’s work as “excellent innovation,” highlighting that reasoning models demand more compute — a boon for Nvidia’s business.

Even Meta's Mark Zuckerberg acknowledged that rising AI competition, including from DeepSeek, will require strategic investment in infrastructure.

A rising force in global AI

As of March, DeepSeek logged over 16.5 million visits, making it the second most-visited AI chatbot app after ChatGPT — although traffic had dipped slightly from February.

Despite being barred in some jurisdictions, DeepSeek has already achieved something few expected: it has forced global tech giants to adapt. Whether by choice or policy, the company is redefining what global AI innovation looks like in an increasingly fractured digital landscape.

Stay tuned to The Horizons Times as we continue to track the disruptive players, political tensions, and frontier technologies reshaping the global AI ecosystem.

Prev Article
Quantum computers could break RSA encryption with fewer than a million qubits
Next Article
Rise and Fall of Tech Unicorns (2010–2025) – Data-Driven Global Study

Comments (0)

    Leave a Comment