DeepSeek AI: What Just Happened, and Why It Matters
The AI landscape just hit refresh. China’s DeepSeek AI has surged to the top of Apple’s App Store, kicking OpenAI’s ChatGPT off its throne as the Top AI App. The hype was real—so real that by January 27th, NVIDIA, king of AI chips, saw its stock dip by 17%. That’s a whopping $593 billion in market value—a record one-day loss for any company.
It’s important to note that this decline was’t solely due to DeepSeek’s win on the App Store. Let’s break it down.
NVIDIA has long been the dominant supplier of AI chips, with its GPUs powering most AI models. The U.S.-China tensions and chip restrictions prevented NVIDIA from selling its most advanced AI chips (like the H100) to China. This pushed Chinese firms like DeepSeek to use older hardware i.e. NVIDIA’s H800 GPUs. Ironically, these old chips contributed to overall lower expenses during the training phase.
DeepSeek’s popularity signals a growing demand for homegrown, cheaper AI models. Safe to say, everyone’s paying attention. But what does this mean for enterprises, solution providers, and the future of AI investments in ASEAN?
Wait, What’s DeepSeek AI?
For those who blinked and missed it, DeepSeek AI is China’s latest large language model (LLM). Developed by Hangzhou-based DeepSeek, this AI contender isn’t just some ChatGPT wannabe. According to its own benchmark tests, DeepSeek V3 outpaces both downloadable, openly available models like Meta’s Llama and “closed” models that can only be accessed through an API, like OpenAI’s GPT-4o. This is a bold flex, considering it’s coming straight out of China’s backyard.
Launched most recently, DeepSeek’s R1 “reasoning” model claims to match Open AI’s 01. Unlike your average LLM, R1 actually fact-checks itself—because, let’s be honest, we’ve all seen AI confidently spout nonsense before. Reasoning models take a few extra seconds (or minutes) longer but the upside is that it tends to be more reliable in domains such as math, science, and physics.
What’s Fuelling the DeepSeek Hype?
Regulatory advantage – With China’s strict AI regulations, DeepSeek AI has home-field support, while Western LLMs like OpenAI struggle with compliance issues.
Localisation – DeepSeek AI is trained in Chinese, making it more aligned with local cultural and linguistic nuances than its Western counterparts.
Growing demand for China-first AI – Enterprises in China are increasingly looking for domestic alternatives to foreign AI tools amid rising geopolitical tensions.
Cost comparisons - While OpenAI reportedly spent millions (estimates put it between $41 million and $100 million) training its GPT-4 model, DeepSeek AI was built on a significantly lower budget (under $ 6 million to train), showcasing China’s ability to create competitive AI models at a fraction of the cost.
While there is currently scrutiny surrounding the costs involved to develop and train DeepSeek’s AI model, AI scientists broadly agree that it marks a positive step for the industry. Yann LeCun, Chief AI Scientist at Meta, writes:
“Deep Seek’s success represented a victory for open-source AI models, not necessarily a win for China over the U.S. To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI,’ you are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones.’”
The AI Gold Rush Is About to Get More Competitive
Key factors at play:
China is investing billions in AI infrastructure to increase self-sufficiency and reduce reliance on U.S. chipmakers like NVIDIA.
New players like DeepSeek are entering the game, meaning enterprises have more choices when it comes to AI models and chips.
The future of AI hardware is shifting, with companies like Huawei developing their own AI chips to rival NVIDIA.
This isn’t NVIDIA’s downfall—far from it. But it’s a reminder that AI dominance is never guaranteed. And for ASEAN enterprises, this is the perfect time to rethink AI strategies.
What This Means for ASEAN Enterprises
As AI capabilities advance, enterprises in ASEAN are making strategic decisions about their AI investments. According to the 2024/2025 AIBP ASEAN Enterprise Innovation Market Overview Report:
78.5% of enterprises see moderate to significant AI investments as crucial to staying competitive.
56% are cautiously exploring AI use cases while focusing on upskilling.
22.5% are taking a bolder approach, building in-house AI teams and custom AI solutions.
DeepSeek AI’s rise is a reminder that AI models are not one-size-fits-all. For ASEAN businesses, this means reassessing their AI stack, compliance strategies, and infrastructure needs. It’s not just about choosing between OpenAI or Google’s Gemini anymore—regional enterprises may soon find China-built AI models offering competitive alternatives tailored to the Asian market.
The AI industry shifts overnight. Literally. Today, it’s DeepSeek AI overtaking ChatGPT; tomorrow, it could be an ASEAN-built AI making headlines.
For enterprises in the region, the message is clear: stay agile, invest wisely, and don’t get too comfortable with today’s AI leaders.
