AI hardware is no longer a speculative bet—it's a battleground, and Etched just signaled it's here to dominate. Three Harvard dropouts launching a chip company in 2022 might’ve sounded naive just a few years ago, especially when betting against GPU giants like Nvidia. But now, after a $300 million Series C at a $10.3 billion valuation—doubling from just seven months prior—it's clear investors aren’t just interested, they’re convinced.
What stands out isn’t the funding alone, but how Etched is reengineering AI inference from the ground up. Most chips treat inference as a monolithic task. Etched splits it intelligently: a dedicated prefill chip optimized for heavy compute at low voltage, and a revolutionary cluster-scale memory system for the memory-hungry decode phase. This isn’t incremental improvement—it’s architectural reinvention. And unlike assumptions, their systems aren’t locked to transformer models. They’re running Mamba, MoE, and more, proving flexibility at scale.
In my experience prepping founders to pitch investors, one thing stands out: traction beats skepticism. Etched’s secret weapon wasn’t just tech—it was private demos. Getting names like Karpathy, Hinton, and Thiel wasn’t about hype; it was about letting elite minds touch and test the hardware. That kind of validation is priceless when you’re challenging the status quo.
And let’s not gloss over the human story. Sleeping on floors, servers in garages, reboot calls to an employee’s wife—this is the raw, unglamorous hustle so many founders endure. Today, they're running a 10MW facility in Milpitas and shipping full rack systems. The scale shift is staggering.
What’s next? Mass production and delivery. The road from prototype to widespread deployment is steep, but if they maintain momentum, Etched could redefine how AI workloads are powered—not just faster, but cheaper and more efficient.
Want to see what the future of AI infrastructure looks like? This is one to watch closely.
Working with investors and entrepreneurs to gain the best ROI possible.
A new service offers uncensored AI models capable of generating exploit code and dangerous biopro...
Language skills are a strategic advantage in business—here’s how to build them with a proven, lin...
What if 'impossible' is just a story you're telling yourself? Robert Irvine breaks down how entre...
A quiet player in AI training data is scaling fast—and taking a stand on ethics in the process.
How Konstantin Sintsov turned rail logistics into a legacy of community and sport
AI is reshaping how Americans manage money — but blind trust could lead to costly mistakes.