DeepSeek, a Chinese AI startup, just dropped a bomb on the AI scene—its AI assistant topped the US Apple App Store.,Trained on Nvidia’s H800 chips for under $6 million, DeepSeek’s model is competing with AI giants who spend billions.,This raises huge questions about US AI dominance and whether export controls on advanced chips are working.,Unlike OpenAI’s closed models, DeepSeek is open-source, letting developers access and tweak it freely.,The AI race just got a whole lot more interesting—so, what happens next?
Wait, Who Is DeepSeek, and Why Is Everyone Talking About It?
Imagine a relatively unknown AI startup dominating Apple’s App Store—in the United States, no less. That’s exactly what DeepSeek just pulled off.
Their AI assistant, built on the DeepSeek-V3 model, blew up overnight, surging to the top of the free app charts. The hype was so intense that cyberattacks took the app down temporarily. Yep, they got too popular, too fast.
But here’s what’s really wild: 💡 DeepSeek built a cutting-edge AI model for under $6 million. 💡 Silicon Valley’s AI giants? They’re spending $100M+ just to train a single model.
DeepSeek isn’t just shaking up the AI world—it’s rewriting the playbook.
Why This Matters: A Direct Challenge to US AI Dominance
DeepSeek’s rise is making a lot of people in Washington nervous.
For years, the US has controlled access to top-tier AI chips, hoping to slow down China’s AI progress. But DeepSeek trained its model using Nvidia’s H800 chips—less powerful than the restricted H100s—and still built an AI that rivals OpenAI and Anthropic. This echoes concerns about the US-China tech war.
This raises a massive question: 👉 If a startup can train world-class AI for a fraction of the cost—without cutting-edge chips—how effective are US export controls, really?
Industry insiders are now rethinking the whole "AI dominance" narrative. If cost-effective AI is possible, the whole game changes. This also highlights the growing importance of AI data scarcity as a bottleneck.
How Does DeepSeek Stack Up Against OpenAI?
Alright, let’s get into the real AI showdown:
Feature,DeepSeek-R1,OpenAI’s o1
Enjoying this? Get more in your inbox.
Weekly AI news & insights from Asia.
Performance,Matches/beats OpenAI’s o1 on math & reasoning tasks,Stronger in creative writing & brainstorming
Cost to Train,$5.6M (yes, million, not billion),Estimated $100M+
Processing Speed,Up to 275 tokens/sec,~65 tokens/sec (o1 Pro)
API Pricing,$0.55 per million tokens (input), $2.19 (output),$15 (input), $60 (output)
Hardware Needs,Runs on consumer-grade GPUs (e.g., 2x Nvidia 4090s),Needs high-end, expensive hardware
Open-Source?,Yes—fully open-source under MIT license,Nope—completely closed
🚀 Bottom line? DeepSeek isn’t just cheaper—it’s faster, open-source, and proving that AI doesn’t have to be a billion-dollar game. This shift could impact how we view the value of data in the AI era.
But… What’s the Catch?
Not everyone’s convinced that DeepSeek is playing fair. A few major concerns have popped up:
⚠️ US Regulators Are Watching: Washington is investigating whether DeepSeek used restricted AI chips—if violations are found, we might see more trade bans.
⚠️ Skepticism Over Costs: Some experts aren’t buying the $6M claim—did they secretly rely on pre-trained models instead?
⚠️ Corporate Blockades: Hundreds of businesses and government agencies have already restricted DeepSeek’s AI, citing security and intellectual property risks.
So… Is This the Beginning of a New AI Era?
DeepSeek’s rise is a wake-up call for the entire AI industry. It proves that:
✅ You don’t need billions to train a competitive AI model. ✅ Restricting hardware access might not stop innovation. ✅ Open-source AI could disrupt the power balance of AI giants.
If a tiny startup can shake up Silicon Valley this much in under two years—what happens next? This disruption could be a precursor to the AI wave shifting to the Global South. For more details on the economic implications of AI, a report by the National Bureau of Economic Research discusses the potential for AI to reduce firm-level costs here.
Your Turn: What Do You Think?
🔹 Is DeepSeek proof that AI development is shifting towards cost efficiency over brute-force spending? 🔹 Will this challenge OpenAI and Google’s AI monopoly, or will regulators shut it down? 🔹 Would you trust an open-source AI over a closed, corporate-controlled model?
Drop your thoughts in the comments! 👇
Want more straight-forward insights on AI in Asia? Subscribe to our newsletter to AIinASIA for the latest AI trends, breakthroughs, and battles that matter. 🚀
Latest Comments (6)
Interesting read, but I'm not entirely convinced just yet. Six million quid is a decent sum, mind you, and while it's a fraction of what the giants spend, true "disruption" means more than just a cheaper model. The real test is adoption and sustained utility, especially in diverse markets like ours where local nuances are key. Let's see if DeepSeek can truly democratise AI beyond the Valley's echo chamber.
Interesting news! It is impressive DeepSeek launched so affordably. However, I do wonder about their long-term sustainability. Silicon Valley has deep pockets and established networks. A lean start up could face an uphill battle, even with a great product. The real test will be how they scale and secure future funding, not just the initial launch cost.
Wow, this is a fascinating read! Six million dollars for an AI that’s genuinely causing a stir is pretty incredible. We always hear about these massive, billion-dollar investments, so this definitely changes the narrative a bit. What I’m really curious about is how DeepSeek plans to sustain its growth and innovation without the deep pockets of the Big Tech players. Will their ‘lean and mean’ approach be enough to keep pace with the huge R&D budgets of the established giants, especially as the AI landscape evolves at such a blistering pace? It’s a real David and Goliath story, lah.
Wah, $6M and already making waves! Curious if their talent acquisition strategy differs from the usual Silicon Valley playbook.
Wah, this is quite a game-changer! Here in Singapore, we're always looking for smart, cost-effective solutions. If DeepSeek can deliver genuine innovation at such a lean budget, it really makes you wonder about the inflated valuations we sometimes see. Very keen to see how this plays out for smaller regional players, potentially leveling the playing field a bit.
Wah, $6M for an AI disrupting giants? That's mad. Makes me wonder if those expensive Silicon Valley guys are just overcharging, eh. We've seen it before.
Leave a Comment