Led · Series A · Aug 2026

Velaura AI develops energy-efficient digital design technology for AI compute, built on patented techniques applied at leading semiconductor process nodes. Its flagship product, Titan Core, is a proprietary chip IP and design platform that the company says delivers a 2x to 4x improvement in performance per watt on the mathematical operations used by AI accelerators without sacrificing performance. The underlying technology has been deployed across more than 30 million ASICs, and Velaura licenses it under an upfront fee plus a royalty tied to the power savings its customers achieve. The company says it is engaged with hyperscalers and chip partners at 3nm and 2nm nodes to integrate the technology into next-generation AI accelerators, and is extending the architecture into physical AI systems including robots, drones, and autonomous machines. Founded in 2022 as Auradine, a bitcoin mining ASIC designer behind the Teraflux miner series, the company rebranded to Velaura AI in March 2026 alongside a pivot to AI silicon, and is headquartered at 3200 Coronado Drive in Santa Clara, California. Financially, Velaura AI announced a $110M Series A in August 2026 at a valuation above $1 billion, led by Seligman Ventures; the articles do not state a cumulative total raised across its earlier rounds under the Auradine name.

Led · Series C · Aug 2026

Lumilens develops photonic connectivity solutions intended to connect thousands or hundreds of thousands of GPUs so they can operate as a single system, addressing the limits of copper-based scale-out networks. The team moved from founding to defining, qualifying and shipping its first product within about two years, and that product is deployed in a hyperscaler’s production AI data centers under a multi-billion-dollar agreement. Lumilens emerged from stealth concurrent with its announced raise of more than $900 million from a syndicate of venture and institutional investors. Management says the funding will support building out infrastructure and developing additional products on its roadmap. The company positions itself at the intersection of optics and AI infrastructure to overcome physical constraints on GPU-scale networking.

Led · Series A · Jul 2026

Smallest.ai develops small, specialized voice models designed to mimic human conversational behavior by listening, thinking and speaking simultaneously for near-zero response lag. The company’s models focus on voice-specific nuances such as diverse accents, dozens of languages and robustness in noisy environments. When queries exceed the small model’s knowledge, Smallest.ai routes the request to a larger foundational LLM and briefly places the user on hold to “research” the issue. The startup targets enterprise customer-support and voice companies and lists RingCentral and Truecaller as customers. Smallest.ai was founded in late 2024 and has raised over $21 million in total funding. The current Series A proceeds are being used to advance its small-model approach and scale commercial adoption.

Led · Series C · Jul 2026

Eliyan develops advanced connectivity technologies including silicon-proven NuLink™ PHYs and NuGear™ chiplet families that target I/O, memory-expansion and power-efficiency challenges in modern compute infrastructure. The company is expanding into optical interconnect to complement its existing electrical interconnect portfolio and to support heterogeneous, co-packaged, and bandwidth-intensive compute fabrics. Eliyan positions its platform to span intra-package electrical interconnects through system-level optical fabrics for scale-up and scale-in AI networks. The company is led by CEO and Co-Founder Ramin Farjadrad and is based in Santa Clara, California. Financially, Eliyan raised $145 million in a Series C at a $1 billion valuation. It plans to use the new capital to accelerate product and interconnect development, broaden ecosystem partnerships, and scale manufacturing.

Participated · Series F · Jul 2026

Founded in 2017 and based in Palo Alto, SambaNova designs and sells AI systems such as the SN40L and the next-generation SN50 for high-performance inference and model serving. The SN40L launched in September 2023 and became available on-premises in November 2023; the SN50 was unveiled in February 2026 and is due to begin shipping to customers in the second half of 2026 with SoftBank as its first deployment partner. SambaNova positions itself as a provider of “premium inference,” claiming the ability to fit multi-trillion-parameter models onto a single rack to run them quickly. The company has commercial wins including JPMorgan Chase, Saudi Aramco, Intel, and other Japanese firms. It has a multi-year partnership with Intel that includes co-developing products and go-to-market efforts, and it says current funding will help scale operations and secure its supply chain to meet strong demand.