In a plot twist worthy of a Marvel movie 🎬, OpenAI just clarified it’s not planning to go all-in on Google’s AI chips—despite rumors swirling this week. The ChatGPT creator confirmed it’s still in the “early testing” phase with Google’s tensor processing units (TPUs) but emphasized Nvidia GPUs and AMD chips remain its go-to power sources. 💻
Why the hesitation? Scaling new hardware is like trying to upgrade your phone’s OS mid-TikTok streak—it takes time, new architecture, and serious software tweaks. Meanwhile, OpenAI’s secret weapon? Its own custom AI chip, set to hit a key production milestone this year. 🔧
Here’s the kicker: OpenAI is using Google Cloud services 🤝 to handle its growing computing needs, teaming up with GPU-focused neocloud firm CoreWeave. Talk about frenemies! Meanwhile, Google’s TPUs—once exclusive to its projects—are now powering rivals like Apple and ex-OpenAI leaders’ startups. The AI arms race just got juicier. 🍿
