PowerLattice Technologies is a technology company.
PowerLattice Technologies has raised $25.0M across 1 funding round.
PowerLattice Technologies has raised $25.0M in total across 1 funding round.
PowerLattice Technologies has raised $25.0M in total across 1 funding round.
PowerLattice Technologies's investors include Celesta.
PowerLattice Technologies is a semiconductor startup developing the industry's first power delivery chiplet for next-generation AI accelerators, addressing the escalating power demands of AI computing by tightly integrating power delivery with compute.[1][2] The company serves data center operators and AI hardware makers, solving the "AI power wall"—high energy loss, heat dissipation, and power constraints that limit GPU/accelerator scaling—by reducing compute power needs by over 50%, doubling performance per watt, and enabling 2X raw performance under fixed data center power budgets.[1][2] Founded in 2023 and emerging from stealth in late 2025 with $25 million in Series A funding (total $31 million raised), PowerLattice shows strong early momentum, backed by Playground Global and Celesta Capital, with board additions from former Intel CEO Pat Gelsinger and Celesta's Dr. Steve Fu.[1][3]
PowerLattice was founded in 2023 by engineering veterans Peng Zou, Gang Ren, and Sujith Dermal, who bring decades of expertise in integrated magnetics, analog ICs, power management, and system design from stints at Qualcomm, NUVIA, and Intel, including a portfolio of issued and pending patents.[1][3] The idea emerged from recognizing the AI power wall as clusters scale to hundreds of thousands of GPUs, where traditional power delivery incurs massive losses over inches of distance; their solution moves voltage regulation directly into the processor package, just hundreds of micrometers from the die.[1][2] Early traction includes stealth development culminating in a $25 million Series A in November 2025, led by Playground Global (with Gelsinger joining the board) and Celesta Capital, validating the tech's potential amid surging AI infrastructure demands.[1]
PowerLattice stands out in AI power delivery through these key advantages:
PowerLattice rides the explosive AI infrastructure trend, where datacenter power budgets and cooling limits cap GPU scaling despite compute demand growth.[1][2][4] Timing is ideal post-2025 AI boom, as hyperscalers face "power walls" from million-GPU clusters; market forces like energy costs, grid constraints, and sustainability mandates favor on-package power innovations over legacy PCB-based delivery.[2][5] By enabling denser, cooler AI racks, PowerLattice influences the ecosystem, accelerating SoC designs from Nvidia/AMD peers and easing datacenter expansions, potentially reshaping power as a bottleneck alongside compute.[1][2]
PowerLattice is poised to capture a slice of the multi-billion AI power management market, with next steps likely including tape-outs, hyperscaler pilots, and Series B funding to scale production amid 2026+ AI capex surges.[1][2] Trends like edge AI proliferation and exascale clusters will amplify demand for its chiplet, while competitors chase similar integration; success hinges on yield ramps and ecosystem adoption. As AI power efficiency becomes table stakes, PowerLattice could evolve from niche innovator to essential enabler, doubling AI's raw output without doubling the energy bill—redefining what's possible in the datacenter era.[1][2]
PowerLattice Technologies has raised $25.0M across 1 funding round. Most recently, it raised $25.0M Series A in November 2025.
| Date | Round | Lead Investors | Other Investors |
|---|---|---|---|
| Nov 1, 2025 | $25.0M Series A | Celesta |