AI Arms Race Volume: 4

Weekly AI Signals — Curated by Builders, for Builders

Every week, we round up the most relevant developments in AI, tech, and the business strategies shaping them. Born from internal conversations among senior engineers, product leads, and AI researchers, this digest simplifies the noise into clear signals. Authored by Brod Justice and Ryan McClure, these summaries are built for real practitioners—people shipping code, scaling platforms, and thinking past the hype.

Volume 4

Key Themes This Week:

• Sovereign AI: Nations racing to develop independent AI capabilities
• AGI Skepticism: Practical applications vs. superintelligence hype
• Cost Efficiency: The diverging economics of different AI applications
• Strategic Control: Browser wars becoming AI distribution battles
• Hardware Politics: National interests impacting technical performance
• Global Pricing: Regional strategies to capture emerging markets

Signal vs. Noise: While the industry continues to obsess over AGI and theoretical capabilities, the real action is happening in national AI strategies, practical applications, and the hardware/distribution layers. The Perplexity-Chrome bid and Deepseek’s hardware challenges show how geopolitics and business strategy are becoming inseparable in AI. Meanwhile, OpenAI’s aggressive pricing in India signals the beginning of global AI pricing wars as major players compete for market share in high-growth regions. The diverging cost structures between practical AI applications and more speculative uses suggest we’re entering a phase where efficiency, focus, and market-specific strategies will matter more than raw capability claims.

1. South Korea Joins Sovereign AI Race While EU Sleeps

China gets it, and now so does Korea. Last week Korea announced a sovereign AI policy with plans to launch their own national AI model. The EU has Mistral, but still needs to wake up or face dependence on the goodwill of others, which in the case of the USA could vanish overnight. This is about more than just technology—it’s about national security and economic independence.

Source: CNBC

2. AGI Obsession Distracts from Real AI Opportunities

As Eric Schmidt notes in the New York Times, the USA may be obsessing far too much about Super Human AI (aka artificial general intelligence, or AGI). The release of GPT-5 this month did not appear to get us any closer to that goal—we remain big skeptics about the AGI hype and believe AGI is many, many years away. The NYT asks “Is the USA bypassing crucial opportunities to use the technology that already exists,” leaving the real gains to China? Incidentally, or perhaps not, Schmidt is an investor in Europe’s Mistral who don’t obsess about AGI, but do produce the AI models most popular with our customers.

Source: New York Times

3. AI Costs: Rising or Falling? It Depends What You’re Building

Many in the AI industry are getting very concerned with what they see as the rapidly increasing costs of AI. So why do we so often talk about how our AI costs are falling? It’s because it depends what you do with it. Doing productive work like we do with ChatBar AI and VSP is becoming more and more efficient. Trying to vibe code complex software applications (that nearly always fail) costs rapidly increasing amounts of compute time (aka “Tokens”). Also, as Florent Daudens of Hugging Face says, “don’t use a cargo ship to deliver a pizza.”

Source: LinkedIn

4. Perplexity’s $34.5B Chrome Bid: Stunt or Strategic Genius?

Perplexity has bid $34.5bn for Google’s Chrome. Some call it a stunt, but it makes sense. Perplexity already has their own browser that is accused of stealing content by many including the BBC, so why do they want to get hold of Google’s Chrome? Well, because whoever has Chrome also has the best signals on what is popular on the Internet and thus what to include in its search results. Losing Chrome would be hugely damaging for Google.

Source: BBC

5. China Forces Deepseek to Use Local Hardware, Model Fails

The Financial Times reports that Deepseek is delaying its AI model roll-out after the Chinese government put pressure on it to use Chinese hardware. It reports that this caused the new AI model to be a failure. Will Deepseek go back to NVIDIA now that Trump has backed down on restrictions? This highlights the geopolitical complexities of AI development and the real-world consequences of hardware nationalism.

Source: Financial Times

6. OpenAI’s $4.6 India Plan: The Global AI Pricing Wars Begin

OpenAI has launched “ChatGPT Go,” a new subscription plan in India priced at just 399 rupees ($4.6) per month—its cheapest offering yet. This aggressive pricing targets India, OpenAI’s second-largest market by user base. The plan offers 10x more message limits, image generations, file uploads, and double the memory compared to the free tier. Sam Altman praised India’s rapid AI adoption, and notably, they’re allowing payment through UPI, India’s local payment framework. This is clearly about market share, not profit margins.

Source: Ground News