DEVIN BLITZER

// AI OPERATING PARTNER · DATA CENTER SERVICES · ELECTRONS TO TOKENS

The money is already in the build. I find it, operate the fix, own the outcome.

Power and AI almost never sit in one person. I'm fluent in both, aimed at the firms that commission and power data centers: Cx, electrical, critical-power. The AI operating layer finds the margin buried in rework, slip, and estimating, proves it on one site, then scales across the platform. Milestone equity in the firm, deal by deal. Never a seat.

Both stacks fluent in power and AI,
the bridge almost no one is
One firm first prove the number on one site,
then scale the platform
Equity in the firm milestone, deal by deal,
never up front, never a seat

// 001 · PROOF

Built on the owner's side of
large-scale infrastructure.

Hyperscalers served

At QTS, Blackstone's largest data-center platform, the AI that runs the program is mine, raw land to live floor. I'm the tip of the spear of the platform's AI.

Most of the value is coordinating the people: the AI directs the crews, kills rework, and clears problems before they cascade, on the systems they already run.

Find the buried money, operate the fix, own the outcome.

Platform I run AI across
QTS · Blackstone's largest data-center platform
Scale
~3GW commissioned · 90+ data centers · 20+ markets · $25B pipeline
My layer
AI across the full lifecycle: site selection → construction → live operations
Runs inside the systems your program already uses
and many more

// 002 · ELECTRONS TO TOKENS

Plug me in where it's leaking.
The whole chain is the edge I bring to any one stage.

One build, seen two ways: the work over time, raw land to live floor, and the power through it, electrons to tokens. I run all of it. Plug me in where the money's worst, prove it there, then expand.

// WHERE YOU PLUG ME IN  ·  THE PROJECT, OVER TIME

  1. 01Pre-development Land, utility, interconnection, permitting, zoning, environmental permit trackers · ERP
  2. 02Design Architects, civil, electrical, mechanical, structural, fire Procore
  3. 03Procurement Switchgear, generators, UPS, chillers, transformers, busway Procore ERP · the master schedule
  4. 04Construction Electrical, mechanical, concrete, steel, low-voltage, controls Procore RFIs
  5. 05Commissioning CxA, electrical contractors, controls vendors, OEM techs CxAlloy
  6. 06Turnover O&M manuals, as-builts, warranties, punch list, training CxAlloy Procore
  7. 07Operations Facilities, DCIM, BMS, EPMS, maintenance vendors Maximo ServiceNow DCIM · CMDB
  8. 08Expansion Live-site crews, MOP / SOP / EOP, outage windows ServiceNow CMMS

plug in anywhere

// THE SAME BUILD  ·  THE POWER, THROUGH IT

Electrons
  1. 01

    Generation & Substation

    Power is the whole game now, not land or capital. The most expensive mistakes lock in before the shovel: the interconnection queue, on-site generation, the substation, the energization date. The deal is won or lost here.

  2. 02

    Critical Power Train

    The most capital and the most risk in the building: the medium-voltage lineup down to the busway and whips at the rack. Six to eight conversions, A-bus and B-bus, each one bleeding efficiency, one fault from dropping a hall. The 2N holds, or it fails at 3am when the utility drops.

    • ASCO
  3. 03

    Commissioning & IST

    Where builds get stuck, and a slipped date costs the most. Commissioning runs Level 1 to the Level 5 blackout test that has to prove the whole facility before service. Broken handoffs burn millions; a three-month slip runs into nine figures. That last test can't be compressed, so the edge is catching drift early. That's what the owner-side AI does.

    • CxAlloy
    • Bluebeam
  4. 04

    Compute & Cooling

    The white space. High-density GPU pods can't power on until the liquid cooling and CDUs are commissioned and leak-tested under load, or six figures of silicon cooks. The compute itself, racked and burned in, not just the shell.

    • Trane
    • CoolIT
  5. 05

    Controls & the AI Layer

    BMS, EPMS, SCADA, DCIM: four control systems that never talked, now one pane of glass. The AI operating layer sits on top and turns the whole chain into one number that moves. Where electrons become tokens.

    • Johnson Controls
Tokens

// 003 · THE MODEL

Even billion-dollar programs bleed money no one can see.
My job is to find it, then make it stop.

  1. 01

    Audit

    I start from the data the program already generates: schedules, commissioning logs, sub and vendor performance. I show you exactly where every dollar is going before I change a thing.

  2. 02

    Access

    Inside the real data (commissioning, operations, maintenance, schedule, P&L), I go deep, surface the buried money and the buried problems, and rank every one by dollar impact.

  3. 03

    Operate

    I don't hand over a report. I deploy the AI that runs the work and the people, in real time, through the tools the field already uses. The number moves: operations tighten first, then the gains compound.

  4. 04

    Outcomes

    The number moves first. Once the savings are real and proven, we structure the deal: equity against defined milestones, sized to the value created. Earned by the outcome, negotiated after the proof, never pitched up front.

// 004 · THE DIFFERENCE

The same operational layer.
None of the baggage.

// LET'S TALK

Rolling up DC-services firms?
Let's prove the number on one.

devin@dblitz.com
LinkedIn Founder, dBlitz · AI Lead, QTS Data Centers Miami Milestone equity in the firm · proven on one site first, never up front, never a seat