AI chip startups Euclyd and Fractile have each closed record funding rounds, according to The Tech Buzz, marking the latest capital infusion into companies building alternatives to Nvidia's dominant AI accelerators.

The two firms join a growing cohort of semiconductor challengers raising money to compete in a market where Nvidia currently captures the overwhelming share of AI training and inference hardware spending. The Tech Buzz did not disclose the round sizes, investors, or valuations for either company.

Who are Euclyd and Fractile?

Both companies position themselves as rivals to Nvidia in AI silicon. Fractile has previously drawn attention for an architecture that aims to accelerate large language model inference by moving computation closer to memory, a design philosophy several startups are pursuing to sidestep the memory-bandwidth bottleneck that constrains conventional GPUs.

Euclyd, the less publicly visible of the two, is likewise building AI inference hardware aimed at the same workloads that today run on Nvidia's data-center product lines.

The rounds are described as records for both companies, signaling that investors continue to reward early-stage silicon startups despite the high cost of competing with an incumbent that pairs silicon with a mature software ecosystem.

Why does the money matter?

Chip development is capital-intensive in a way most software ventures are not. A credible AI accelerator requires:

• Multi-year architecture and RTL design work before first silicon
• Expensive foundry and packaging allocations at advanced nodes
• A software stack that can attract developers away from Nvidia's CUDA
• Sustained funding to survive long design cycles between revenue events

Record rounds give Euclyd and Fractile the runway to reach working silicon and early customer deployments — the two milestones that separate funded contenders from credible ones in the inference-acceleration segment.

What does this say about the competitive picture?

Investor appetite for Nvidia alternatives has remained strong even as Nvidia's own data-center revenue has climbed. Backers are betting that the economics of large-scale AI inference — where power efficiency and cost per token matter more than raw training throughput — leave room for specialized architectures to take share.

That bet carries risk. Several well-funded AI chip challengers have struggled to convert architectural advantages into volume orders, largely because hyperscalers weigh software maturity and ecosystem lock-in alongside performance. Both startups will now have the capital to test whether their approaches clear that bar.

With fresh funding secured, the next measurable milestones for Euclyd and Fractile will be product announcements, silicon availability, and named customers — the signals that will show whether the record rounds translate into commercial traction against Nvidia.