Niv-AI is a Tel Aviv-based startup that builds software to manage power spikes and optimize electricity use in AI data centers. [bydj2h] [v11e6j]

Company Overview

  • Founded: May 2025 by CEO Tomer Timor and CTO Edward Kizis.
  • Funding: Raised $12 million in a seed round in March 2026.
  • Investors: Led by Glilot Capital Partners, with Grove Ventures , Arc VC , Encoded VC , Leap Forward Ventures , and Aurora Capital Partners . [v11e6j]

Core Technology

  • The Problem: High-intensity GPU workloads create sharp, millisecond-level power spikes. Data centers waste up to 30% of their power capacity maintaining safety buffers against these surges. [v11e6j]
  • The Solution: Niv-AI uses high-frequency sensors to capture the "electrical fingerprint" of AI workloads. [es8o6t]
  • The Result: Its AI models predict power shifts and balance workloads in real time. This unlocks trapped electrical capacity without adding bulky physical hardware or slowing down GPUs. [es8o6t]

Value Proposition & Features

Niv-AI is an Israeli AI power-orchestration company that analyzes AI-workload fingerprints and dynamically schedules compute to synchronize demand with the electrical grid, reducing instantaneous power spikes. [E1] Its stated goal is to help data centers unlock stranded power capacity without adding hardware while increasing tokens per watt for AI training and inference. [E1]
Core capabilities include:
  • AI workload fingerprinting: Identifies the power-demand characteristics of AI workloads. [E1]
  • Dynamic Workload Orchestration : Adjusts compute scheduling in response to workload and grid conditions. [E1]
  • Power-spike reduction: Coordinates demand to eliminate instantaneous electrical-load spikes. [E1]
  • Stranded-capacity utilization: Enables data centers to use available power capacity that would otherwise remain inaccessible because of peak-demand constraints. [E1]
  • Hardware-free capacity expansion: Pursues greater usable capacity without requiring additional servers or electrical hardware. [E1]
  • Training and inference optimization: Targets higher tokens-per-watt efficiency across both AI training and inference. [E1]
  • Data-center Grid Synchronization : Connects AI-compute operations with electrical-grid availability. [E1]
The company’s positioning fits the emerging AI-factory and data-center operations market, alongside organizations pursuing hyperscale and neocloud infrastructure strategies. Datacenter Operations Systems [E1]

Screenshots

No three official public screenshots were identified in the available search results.

Product Roadmap / Announcements

As of October 8, 2026 , no reliable public roadmap or dated product announcements from the preceding six months were found in the available search results.

Recent Developments

No reliable news coverage or independently verified developments from the past 90 days were found in the available search results. A company-directory result describes Niv-AI as founded in 2025 and based in Tel Aviv, Israel. [o3n8hn] The directory also lists Aurora Capital Partners and five additional investors, but it does not provide a sufficiently detailed financing announcement to verify the round structure or investor participation independently. [o3n8hn]

History and Origin Story

Niv-AI appears to be a Tel Aviv-based startup founded in 2025, focused on coordinating AI-compute demand with electrical-grid constraints. [o3n8hn] The available company material identifies the product thesis—workload-aware power orchestration—but does not reliably identify the founders, founding circumstances, or major historical milestones. [E1]

Fundraising History

Round Date Amount Lead investor
Seed 2026-03-01 USD 12.0M No reliable source found
Total — USD 12.0M reported —
The retrieved company material reports total funding of USD 12 million from a seed round dated March 1, 2026. [E1] A separate directory lists Aurora Capital Partners and five additional investors but does not establish which investor led the round. [o3n8hn]

Market Sizing

Category, Market Size, and Category Growth

Niv-AI is best categorized as AI data-center infrastructure software , specifically workload orchestration, power management, and data-center energy optimization. Its stated product addresses the growing operational constraint created when AI training and inference produce large, rapidly varying electricity loads. [E1]

Competitive Landscape

Who it's for, who it's not for

Niv-AI is aimed at AI data-center operators, hyperscalers, neocloud providers, and other organizations whose training or inference workloads are constrained by electrical-capacity peaks. Its value proposition is strongest where existing infrastructure has available average capacity but cannot safely support AI workloads’ instantaneous demand spikes. [E1]
It is not evidently aimed at general-purpose cloud-cost management, model development tooling, or organizations without control over compute scheduling and facility power operations. No public source confirms a broader target market.

Viable Alternatives

  • NVIDIA DSX ecosystem: Relevant for data-center infrastructure coordination and AI-factory operations, though the available results do not establish feature-level equivalence with Niv-AI. Datacenter Operations Systems
  • Traditional Data Center Infrastructure Management Systems : Potential alternatives for monitoring and operational control, but no specific vendor was reliably identified in the available search results.
  • Cloud Workload Orchestrators : Potential substitutes for compute-placement and timing decisions, although they may not optimize against electrical-grid constraints.
  • On-site power and electrical upgrades: A hardware-based alternative to software orchestration when capacity constraints cannot be addressed through scheduling alone.

Competitor Table

Competitor Description
NVIDIA DSX AI-Factory and Data-Center Operations ecosystem referenced in the available market context; direct parity with Niv-AI was not established.
Traditional data-center infrastructure-management platforms Monitor and control facility operations, but the available sources do not identify a specific platform with Niv-AI’s stated grid-synchronization proposition.
Cloud workload schedulers Schedule or place compute workloads, but the available results do not show that they coordinate demand with electrical-grid constraints.
Electrical-capacity upgrades Add physical power capacity rather than optimizing existing capacity through workload orchestration.

Sources

Sources (retrieved)

[E1] Exa.ai API response for data on Niv-AI

[E2] Exa.ai API response for data on nivii.ai

[E3] Exa.ai API response for data on NIVA

[E4] Exa.ai API response for data on Nivtrix Inc.

[E5] Exa.ai API response for data on Nivo

[E6] Exa.ai API response for data on Wiv.ai