Artificial intelligence news.
Straight from the original announcements.

Ampt unveils battery converter for higher-power 800 VDC AI data centers

November 11, 2026, 12:00By the AI News Bites Editorial TeamSource: PR Newswire
Ampt unveils battery converter for higher-power 800 VDC AI data centers
Image: PR Newswire

Ampt has introduced a battery converter designed to help 800 VDC AI data centers run more compute from the same grid connection. The rack-mountable unit responds to GPU power changes in less than 1 millisecond, allowing batteries to handle short demand peaks while the grid supplies a steadier average load.

Key points
  • The converter responds to GPU load changes in less than 1 millisecond.
  • Each 2U unit delivers 80 kW, or 100 A, at regulated 800 VDC.
  • Units can be paralleled from one rack to an entire GPU cluster.
  • Ampt says peak efficiency reaches up to 99.2%.
  • General availability is planned for Q3 2027 after partner pilots.

Battery storage absorbs fast GPU demand swings

Ampt's Battery Converter delivers regulated 800 VDC from a DC bus shared by battery storage and grid or on-site power. When GPU demand changes, the converter enables the battery to supply or absorb the difference, while the grid or on-site generation experiences a more predictable load.

AI workloads can move a rack's power draw between 30% and 100% within milliseconds, according to the source. That creates a mismatch between the GPUs' rapidly changing needs and the grid's preference for a steady load. NVIDIA's white paper says energy storage needs to fill the gap between those requirements.

By using the battery for short peaks, operators can run GPUs without power capping and size the grid connection for average load rather than peak load, according to Ampt. The company says this can reduce upstream equipment costs and help sites secure grid access sooner.

80 kW units scale from one rack to a GPU cluster

Each Battery Converter occupies 2U of rack space and delivers 80 kW, or 100 A. Units can be connected in parallel from a single rack to a full row or an entire GPU cluster. Adding one unit provides N+1 redundancy, allowing full power to continue if a unit goes offline.

The battery, grid and any on-site generation feed a common DC bus. The grid connects through a medium-voltage transformer and power conversion system, while batteries can operate up to 1500 VDC. The bus runs at 1100 to 1500 VDC, depending on the battery's state of charge, before the converter supplies stable 800 VDC to GPU racks.

Ampt says the converter reaches up to 99.2% peak efficiency. Because stored energy stays DC, the design avoids DC-to-AC-to-DC conversions used by AC-coupled storage. Higher voltage can also carry the same power at lower current, reducing conductor losses or allowing smaller conductors.

Partner pilots are underway ahead of Q3 2027 availability

The Battery Converter is currently in partner pilots for data centers, with general availability planned for Q3 2027. Sites with existing or planned solar can connect PV systems to the same battery DC bus through Ampt String Optimizers, without an additional inverter or transformer, according to the company.

Ampt CEO Aaron Gomolak said, “In AI factories, GPU loads swing in milliseconds, so storage has to work every minute of the day.” He said the converter gives the grid a steady load and GPUs fast-responding power, which Ampt believes can support reliable compute and a faster path to bringing new AI capacity online.

The product is aimed at data centers moving toward 800 VDC architectures. Those systems are projected to support 1 MW IT racks and improve end-to-end power efficiency by up to 5% compared with 54 V systems, according to NVIDIA.

Related news

Meta Introduces TLX, a Hardware-Native Compiler for GPU Workloads
For developersNovember 10, 2026, 17:59

Meta Introduces TLX, a Hardware-Native Compiler for GPU Workloads

Meta has detailed Triton Low-Level Extensions, or TLX, an extension to Triton designed to improve GPU performance in production machine learning workloads. TLX gives…

NVIDIA Snapshot restores GPU workloads in seconds on Kubernetes
Chips and infrastructureNovember 10, 2026, 16:00

NVIDIA Snapshot restores GPU workloads in seconds on Kubernetes

NVIDIA has introduced Snapshot, an open-source checkpoint and restore system for GPU workloads on Kubernetes that can bring initialized inference replicas online in…

Business
BusinessNovember 4, 2026, 20:29

Fermaca becomes Mexico's first NVIDIA Cloud Partner, unveils AI factory

Fermaca has been designated the first NVIDIA Cloud Partner based in Mexico and the first of its kind in the region, and has unveiled an AI factory platform aimed at…