Badal Khatri

Everyone’s chasing bigger models. But the real shift? Smarter, smaller, and local. Instead of one giant AI doing everything, we’re moving toward: Tiny specialist models Running directly on your device Coordinated by a simple router Using the cloud only when needed This week, GLM-OCR proved the point. Just 0.9B parameters yet outperforming models like Gemini in document understanding. No brute force. Just better design: Splits documents into regions Processes in parallel Generates faster, cleaner results And it runs locally with Ollama. That’s the direction. Not bigger AI. Better architecture.

Claude wasn’t permitted to write outside the workspacebut it attempted to anyway So it generated a Python script and ran it through bash, effectively bypassing the restrictions and modifying the file by hacking the permissions.

AI, chips, cloud, geopolitics, trillion-dollar bets… The last 24 hours just revealed where tech is actually heading. Here’s the real signal: Intel’s next-gen chips show laptops are becoming AI workstations. This isn’t about faster devices anymore it’s about who controls on-device intelligence for billions of users. At the same time, a reported strike on a cloud data center changes the rules: Cloud is no longer neutral infrastructure. It’s now geopolitical territory. That means the future isn’t just faster or smarter it’s more fragmented and more strategic. China is already moving there. Running advanced AI models on domestic chips = independence from US tech. This is the beginning of parallel AI ecosystems: US stack. China stack. Regional stacks. Less interoperability. More competition. Now follow the money: Microsoft is putting $10B into AI infrastructure in Japan. IT giants are restructuring around AI. And SpaceX might be heading toward a $1T IPO. This isn’t hype. It’s global infrastructure being rebuilt in real time. Zoom out and the pattern is clear: Chips are localizing Cloud is militarizing AI is fragmenting Capital is scaling This isn’t a trend. It’s a reset.

Building something cool.

@ZayleeY10909 Absolutely

@BadalXAI One of the most overlooked aspects of cryptocurrency economics is the trade-off between efficiency, security, and cost in real-world applications.

AI isn’t slowing down. It’s breaking things power grids, chip dominance, cybersecurity, and reliability. The last 24 hours prove one thing: The AI race just entered a more dangerous phase.

13/🧵 The builders who win won’t chase trends. They’ll understand systems: AI + hardware + policy + distribution That’s the new stack. Bookmark this you’ll want to revisit it in 6 months.

12/🧵 Here’s the real pattern: AI → demands compute Compute → depends on geopolitics Geopolitics → reshapes supply chains Supply chains → create new winners Everything is connected.

In the last 24 hours, tech quietly shifted. AI monetization. Supply chain wars. Drone logistics. Moon missions. Most people missed it. You shouldn’t.

13/🧵 Bookmark this. You’ll want to come back when these “small updates” turn into category-defining companies.

12/🧵 Zoom out: AI is becoming invisible Healthcare is becoming predictive Crypto is becoming regulated Hardware is becoming critical Different stories. Same direction: consolidation of power.

Where are you currently in the AI stack?