How OpenAI's Push to Ban DeepSeek Mirrors the Playbook Used Against Huawei

OpenAI recently proposed restrictions on models like DeepSeek, citing "security" concerns. But haven't we seen this playbook before? A few years ago, Huawei was the world's leading smartphone brand by market share. However, when the U.S. found it difficult to compete, Huawei was sanctioned under the guise of "national security." Overnight, Huawei lost access to critical technologies like advanced semiconductors and Google Play Services, causing its market share to plummet.

Today, we're witnessing a similar narrative unfold with DeepSeek AI. This Chinese AI lab has shaken the global tech industry, with companies like NVIDIA losing billions in market value and the tech-heavy Nasdaq-100 experiencing significant declines. What's remarkable is that DeepSeek achieved this with far fewer resources and less computational power, especially considering that China is already restricted from accessing the latest NVIDIA chips.

DeepSeek's cost-efficiency and performance have raised serious questions about the effectiveness of established players like OpenAI and Anthropic. Instead of innovating to compete, OpenAI's response seems to be an attempt to ban DeepSeek, labeling it as a "state-controlled" and "PRC-produced" entity. However, these accusations lack clarity.

It's unclear whether OpenAI's references to "Deep Seek models" are meant to target DeepSeek's API or its open-source models or both. What is clear, however, is that DeepSeek's open models do not include mechanisms that would allow the Chinese government to access user data. In fact, these models are hosted on infrastructure provided by reputable U.S. companies like Microsoft, Perplexity, and Amazon. Given this, the accusation that DeepSeek poses a security threat seems baseless and raises more questions than it answers.

Moreover, OpenAI has blamed DeepSeek for using model distillation techniques on OpenAI's models. This is ironic, given that OpenAI itself trained its models on vast amounts of internet data, often without explicit permission. If OpenAI can leverage internet data to build its models, why shouldn't others use OpenAI models to refine theirs? This double standard raises questions about the true motivation behind these accusations.

History, however, has shown that such tactics rarely succeed in the long term. Take Huawei as an example. After being sanctioned, the company didn't collapse, instead, it adapted. Despite losing access to critical technologies like advanced semiconductors and Google Play Services, Huawei developed its own operating system, HarmonyOS, and built an entirely new ecosystem. Even more impressively, they designed the 7nm Kirin 9000s 5G chip, which powers the Mate 60 series and shocked the global tech industry. While their chips may not yet match the performance of Apple's or Qualcomm's latest offerings, Huawei's ability to innovate under pressure demonstrates resilience and self-sufficiency.

The lesson here is clear: banning competitors through unsubstantiated claims is not a sustainable or effective strategy. Instead of resorting to such tactics, the tech industry should focus on competing through innovation, skill, and collaboration. OpenAI, as a leader in AI, has the resources and talent to push the boundaries of what's possible. Rather than attempting to sideline competitors like DeepSeek, it should strive to outperform them by building better and efficient models.

The attempt to stifle competition will only backfire and push rivals to become more self-reliant and innovative. Let's hope the industry learns from these lessons and chooses a path of fair competition that benefits everyone and accelerates global progress.