Etched, an ambitious artificial intelligence hardware startup founded by three Harvard dropouts, has reached a significant milestone with a new valuation of $10.3 billion. The company has successfully attracted major backing from high-profile investors, signaling strong market confidence in its alternative approach to AI processing.
A Specialized Approach to Inference
Unlike industry leaders that rely heavily on General Purpose Graphics Processing Units (GPUs), Etched has developed custom-designed chips and memory components specifically optimized for AI model inference. The company claims its hardware can significantly increase processing speeds for any AI model without the need for traditional GPU clusters.
By narrowing the focus to inference—the phase where a trained AI model processes new data to provide answers—Etched aims to deliver superior efficiency and performance compared to versatile but power-hungry general-purpose hardware. This specialized architecture could potentially lower the cost of deploying large-scale AI applications across various industries.
