Back to briefings
Download PDF

February 18, 2026

Global

Full-stack cybersecurity and air-gapped cloud adoption by governments AI scaling constrained by infrastructure and energy limits Silicon photonics/DWDM co-packaged optics moving toward high-volume production Power availability and modular design remain top data centre priorities GPU-as-a-Service and data sovereignty shaping African AI infrastructure demand Enterprise consumption models (APEX) supporting low-latency AI/data workflows

Top news (3)

  1. Cybersecurity focus is moving “full‑stack” as AI raises the threat level — At the Munich Security Conference, Google called for a collaborative, end‑to‑end approach and highlighted adoption of Google Distributed Cloud (Air‑Gapped) by NATO, the U.S., Germany, the UK and Australia. See: Google calls for full-stack cybersecurity at Munich Security Conference.

  2. Compute and energy constraints are becoming a visible limiter on AI scaling — A January 2026 “AI Frontiers” digest from Ukraine’s cybersecurity coordination bodies flags rising focus on efficiency/monetization, intensifying US–China competition in semiconductors/compute infrastructure, and infrastructure + energy constraints as key bottlenecks. See: AI Frontiers: January 2026 review of key AI events.

  3. AI networking supply chain is pushing silicon photonics toward high‑volume DWDM optics — Tower Semiconductor and Scintil Photonics announced availability of LEAF Light™, a heterogeneously integrated DWDM laser source validated on Tower’s silicon photonics production lines, targeting DWDM co‑packaged optics for hyperscale AI networking, with a stated pathway from evaluation to “millions of units per month.” See: Tower and Scintil launch DWDM laser sources for AI.


Key deals, projects and corporate moves

Global

  • Enchanted Rock leadership hire (go‑to‑market for data centre/critical infrastructure)

    • Enchanted Rock hires Niki Herr as Chief Marketing Officer: Niki Herr appointed CMO to lead integrated marketing, brand and communications as the company scales across data centers, critical infrastructure, and commercial/industrial markets.
    • Prior roles cited include leadership positions at Interpublic Group agencies, Gensler, and Stem, Inc.
  • Enterprise data + AI operations reference architecture (consumption model)

    • McLaren turns 1.5 TB race data into competitive edge: McLaren processes ~1.5 TB of race‑weekend telemetry/video/simulation data using Dell PowerStore + PowerScale, consumed via Dell APEX and delivered as part of the Dell AI Factory, paired with Dell PowerEdge servers.
    • Emphasis is on low‑latency access for simulations and scalable storage for unstructured datasets—a relevant pattern for AI‑driven workloads in colocation/private DC environments.

Power, grid and interconnection highlights

No specific new grid connections, interconnection awards, PPAs, or substation/transmission projects were detailed in today’s stories.


Policy, regulation and security

Europe / transatlantic security

  • Full‑stack cybersecurity and sovereign/air‑gapped deployments highlighted
    • Google calls for full-stack cybersecurity at Munich Security Conference: Google urged a “full‑stack” collaborative approach; warned of AI‑enabled adversary tactics (via Google Threat Intelligence); and highlighted Google Distributed Cloud (Air‑Gapped) usage by NATO and multiple governments.
    • Google also opened applications for the Google for Startups Gemini Startup Forum: Cybersecurity.

Ukraine / global competition framing

  • AI infrastructure geopolitics and constraints

Regional watch: Africa (AI demand + local capacity)

  • GPU‑as‑a‑Service positioned as a local-data-centre growth lever
    • GPU-as-a-Service powers African AI via local data centres: Cassava AI authors argue GPUaaS delivered from African data centres can improve cost‑efficiency and data sovereignty, enabling local development and deployment.
    • Partnerships cited include Cassava Technologies’ work with NVIDIA as a model of combining global expertise with local infrastructure.

Technology & supply chain (AI networks and optics)

  • Silicon photonics for hyperscale AI networks (DWDM co‑packaged optics)
    • Tower and Scintil launch DWDM laser sources for AI: LEAF Light™ described as the first heterogeneously integrated DWDM laser source using Scintil’s SHIP™ tech, validated on Tower’s production lines.
    • Collaboration targets DWDM co‑packaged optics and cites an AI networking market opportunity of $200B by 2030.

2-line close

Execution risk is increasingly concentrated in (1) cybersecurity posture across the stack and (2) physical constraints around compute and energy. Supply-chain innovation in high‑volume AI networking optics is accelerating alongside demand for more localized GPU capacity.

Subscribe to Data Centres Briefings

Get AI-powered briefings delivered to your inbox

Region