FuriosaAI raises $620M growth round at $2.9B valuation for efficient inference chips
FuriosaAI raised $620M in a growth round at a $2.9B valuation and is scaling its inference chips to 40,000 units by 2027.
FuriosaAI has raised a $620 million growth round at a $2.9 billion valuation. It is scaling its energy-efficient inference chips to 40K units by 2027. The figures come from a one-line entry in a newsletter's deals roundup, which describes the company as an energy-efficient inference chip maker whose chips use 50% less power than NVIDIA GPUs. The roundup names National Growth Fund (Korea), Cisco Investments and NeoFort Holdings as investors.
The 50% figure is a comparison with NVIDIA GPUs, but the entry does not say which GPUs, which workloads or which measurement was used, so the claim is only as specific as that sentence. Inference chips run trained models to produce answers rather than training them.
The deal entry also does not say when the round closed, what the company will use the new capital for, or how many of the 40K units are already committed. The valuation and round size are stated as fact in the entry, but the source is a secondhand summary rather than a filing or announcement from the company.
What changed
FuriosaAI raised a $620 million growth round at a $2.9 billion valuation and plans to scale its energy-efficient inference chips to 40K units by 2027.
Type: Business · Applies to: energy-efficient inference chips
Why it matters
Inference power draw and cost per token are becoming buying criteria, so GPU alternatives deserve evaluation for production serving.
What you can do
Teams planning inference capacity could add energy per token and power draw to their vendor comparisons.
Original source
We have no link to the original source for this story yet.
Where we saw it
The AI Report
AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works