China ditches Nvidia's Chips?

Plus, A new ‘AGI’ Test humiliates top AI models.

AuthorAuthor

Tak Lo & Daniel Akpobare
March 25, 2025

Hello and welcome to the Automated, your AI tour guide.

Just when we thought DeepSeek’s feat might be a one-off and AI development would remain locked behind a few money-burning giants, China’s Ant Group proved otherwise.

Not only did they make AI development cheaper, but they did it without relying on Nvidia’s chips!

Here’s what we have for you today:

  • 🤯 Ant Group’s Cost-Cutting Move Away from Nvidia.

  • 🤖 A New ‘AGI’ Test Just Humiliated Top AI Models—Humans Still Reign Supreme.

  • 🪪 How to use canva AI to create business cards.

  • 🎮 How devs and laymen alike are using AI to create apps and games.

  • 🤖 ChatGPT Prompt Of The Day: Turn customer feedback into a competitive edge.

🤯 Ant Group’s Cost-Cutting Move Away from Nvidia.

AI is crazy powerful, but let’s be real—it’s also ridiculously expensive to develop.

Training these advanced models burns through computing power (and cash) like wildfire, which is why even giants like OpenAI are still figuring out how to actually turn a profit.

Then comes Ant Group, which just pulled off something huge: training an AI model without Nvidia’s top-tier (and expensive) chips—using homegrown Chinese semiconductors instead.

If you know anything about AI, you know Nvidia’s GPUs are basically the gold standard—fast, powerful, and eye-wateringly pricey.

For years, if you wanted to train cutting-edge AI, you had no choice but to pay their premium prices (assuming you could even get their GPUs, thanks to supply chain chaos).

But Ant Group (backed by Alibaba founder Jack Ma) just flipped the script.

Using alternative chips from Alibaba (their in-house AI hardware) and Huawei (Ascend series chips), they managed to hit performance levels on par with Nvidia’s H800—while slashing costs by 20%.

That’s not just a win—it’s a wake-up call for the entire AI industry. If AI development is a financial black hole, this might just be the first real blueprint for digging out.

And this isn’t just about saving money—it’s about survival.

With U.S. sanctions making it harder for China to access Nvidia’s best chips, companies have been scrambling for alternatives. And Ant Group’s success? Perfect timing.

Now, while details on the exact chips used remain under wraps—and while the company still relies on Nvidia for some AI development tasks—this breakthrough proves one thing:

Ant Group (and possibly China as a whole) is shifting toward domestically produced chips, with a little help from AMD, rather than relying on Nvidia or Western suppliers. 

If this scales, it could reshape the entire AI hardware game—turning these alternatives from mere backups into real contenders.

And let’s not forget, China’s already making waves in AI.

Remember DeepSeek AI, the model that outperformed GPT-4 on certain benchmarks? Now Ant Group is proving that AI breakthroughs don’t have to come with an Nvidia-sized price tag.

The big question is: Can these alternative chips keep up with Nvidia’s ecosystem and long-term innovation? Or is this just a temporary workaround?

Nvidia’s ecosystem is still the most advanced, and not every AI company can just switch to Chinese-made hardware overnight.

Either way, one thing’s clear—AI’s biggest challenge (its insane cost) just got a little less intimidating.

And if this trend continues? Well, AI development might finally stop feeling like an exclusive club for companies willing to burn mountains of cash.

Want to find out more? Click here.

10x Your Outbound With Our AI BDR

Scaling fast but need more support? Our AI BDR Ava enables you to grow your team without increasing headcount.

Ava operates within the Artisan platform, which consolidates every tool you need for outbound:

  • 300M+ High-Quality B2B Prospects, including E-Commerce and Local Business Leads

  • Automated Lead Enrichment With 10+ Data Sources

  • Full Email Deliverability Management

  • Multi-Channel Outreach Across Email & LinkedIn

  • Human-Level Personalization

 🤖 A New ‘AGI’ Test Just Humiliated Top AI Models—Humans Still Reign Supreme.

AI is getting smarter, but how smart is it, really? 

A new test just exposed how far even the most powerful AI models are from true intelligence—and the results are brutal.

The Arc Prize Foundation, co-founded by AI researcher François Chollet, has introduced ARC-AGI-2, a puzzle-based intelligence test designed to measure an AI model’s ability to solve brand-new problems.

Spoiler alert: most AI models flunked hard.

🚨 Here’s how they scored:

  • OpenAI’s o1-pro and DeepSeek’s R1 (reasoning models): 1% - 1.3%

  • GPT-4.5, Claude 3.7 Sonnet, and Gemini 2.0 Flash (non-reasoning models): ~1%

  • Humans (average score from a test panel): 60% 🎯

The test is brutal because it forces AI to recognize complex visual patterns in grids of colored squares and then predict the correct answer without relying on training data.

Chollet noted that unlike ARC-AGI-1, brute force computing power won’t save these models anymore.

This matters because AI has been “cheating intelligence tests” by throwing enormous amounts of computation at the problem.

ARC-AGI-2 changes the game by demanding efficiency, not just raw power. AI now has to learn on the fly, just like humans do.

OpenAI’s o3 (low) model, which dominated the previous test (ARC-AGI-1) with a 75.7% score, got wrecked on the new version—managing just 4% while burning $200 of computing power per task. Ouch. 😬

The AI industry has been hungry for new unsaturated tests that better reflect real-world intelligence. Hugging Face co-founder Thomas Wolf recently pointed out that AI still lacks tests that properly measure creativity and adaptability.

Meanwhile, The Arc Prize Foundation is offering a 2025 challenge in a new Arc Prize contest: Developers must get 85% accuracy on ARC-AGI-2 while spending only $0.42 per task. That’s 200x cheaper than OpenAI’s o3 attempt.

The race for true Artificial General Intelligence (AGI) is on—but for now, humans still have the edge. 🏆

Click here to explore more details on this exciting win for humans.

 🧱Around The AI Block

  • Descript: AI-powered audio and video editing tool with transcription, overdubbing, and screen recording.

  • Voiser: AI text-to-speech and speech-to-text tool for generating natural-sounding voiceovers.

  • REimagine Home: AI interior design tool that redesigns and stages rooms from photos.

  • Typeface AI: AI content creation tool for businesses, enabling personalized marketing and branding.

  • Deep Dream Generator: AI-powered image generator that creates dreamlike art from photos and text.

🤖ChatGPT Prompt Of The Day: Turn customer feedback into a competitive edge.

Your customers are constantly giving you valuable insights—it’s just a matter of listening. Their feedback highlights what’s working, what’s not, and where opportunities for improvement lie. By analyzing reviews, testimonials, and support tickets, you can uncover patterns, refine your offerings, and gain a competitive edge.

With ChatGPT-4, you can turn raw feedback into actionable strategies that drive customer satisfaction and business growth.

Here’s a prompt to help you make the most of customer insights: 

Here are recent customer reviews and feedback. What common themes do you see? How could we use this to improve our product/service? [Paste or upload the information].

We've Compiled a List of Over 100 ChatGPT Power Prompts.

This should help streamline your interactions with ChatGPT and get the results you need more efficiently.

Best of all, It's free!

That's all we've got for you today.

Did you like today's content? We'd love to hear from you! Please share your thoughts on our content below👇

What do you think of today's email?

Login or Subscribe to participate in polls.

Your feedback means a lot to us and helps improve the quality of our newsletter.

Reply

or to participate.