Hyperscaling the Trades: Building the Workforce Behind AI Infrastructure

Executive summary 

AI has moved the data center bottleneck from land to power, and from power to people. Salute, which works with 80% of the world’s data center operators (Salute), expects to add 10,000 net new engineers in the next three years. Classroom courses, PDF binders and manually updated procedures cannot scale at that pace. 

In Austin, Sanchack and Henderson described the model replacing them: veteran talent pipelines, immersive 3D and XR training, and living procedures pushed to every technician at once. This paper summarizes what they said and what infrastructure leaders can take from it. 

1. The talent math has changed

When Sanchack joined Salute in 2023, the company had a couple hundred engineers. It expects to end 2026 with about 3,500. The next step is larger still. 

“It pales in comparison to the pipeline requirement in front of us, where we have another 10,000 net new engineers, we need to bring on board in the next three years.” — Erich Sanchack 

Hiring alone does not close that gap. Sanchack said talent acquisition and training had to scale “logarithmically better” than current practice, and that Salute could not do it alone. The answer was a set of partners working together: a veteran training pipeline, a training technology platform, and Salute’s own operating expertise. 

2. AI moved the constraint from land to power, and from power to people

In the cloud build-out of the late 2010s, land and location drove site selection. AI changed that. Sanchack described hyperscaler conversations now centered on power, power density, and keeping high-density equipment cool. 

That shift brought liquid-to-chip cooling into the mainstream. These closed-loop systems limit water use and community impact, but they demand far more of the engineers who run them. The hyperscalers, Sanchack said, will focus on power and the framework; they need a partner to handle “the people and the systems and controls and the infrastructure support” that keep that density cool. 

The skills baseline is rising at the same time the headcount is. New hires must learn today’s standard and the next generation’s standard at once. 

3. Why the traditional training model breaks

Four pressures came up repeatedly in the conversation. 

Pressure	What the interview described	What it demands of training
Scale	10,000 net new engineers in three years	Proficiency in far less time per hire
Static procedures	Updating SOPs and MOPs across 300+ data centers and about 20 GW is “a long manual process”	Procedures that update everywhere at once
Geographic sprawl	AI training campuses sit in remote areas; inference sites sit near users	The same operating standard at every site, delivered remotely
Local regulation	OSHA rules do not carry over to Europe or Canada; a Texas engineer moving to Copenhagen faces different requirements	Localized, regionalized content per technician

The common thread is that a procedure written once, printed, and taught in one central classroom goes stale before it reaches the floor. 

4. A new model: living, immersive procedures

Salute found BILT through its partner UHP, which transitions active-duty service members and veterans into trade roles such as electricians, plumbers, HVAC technicians, and welders. When Salute asked how UHP trained at volume, the answer was the BILT platform. 

Salute concluded the technology was more than a learning tool. It will sit under the company’s standard operating procedures (SOPs), maintenance operating procedures (MOPs) and emergency operating procedures (EOPs). The partnership covers three disciplines: electrical and power, fiber and networking, and HVAC (Nathan Henderson, LinkedIn). 

Three features matter most. 

Retention. Sanchack, who holds an electrical engineering degree from Penn State, tried a BILT lockout/tagout and circuit procedure himself using the XR immersive version of BILT in Apple Vision Pro. 

“It reminded me quite a bit of Tony Stark… He would be inside the system, and he would be moving things with his hands. That was exactly my experience when I used it, and I still remember it.” — Erich Sanchack 

Speed to proficiency. Henderson framed BILT’s goal for Salute as making a new technician 90% proficient in a third of the time, then continuing to add skills as equipment changes. 

Instant updates. When a safety or maintenance procedure changes, BILT pushes it to every technician’s device at once. Sanchack called the value hard to describe “because it’s not something that just exists today.” Because content can be localized, the same push can carry region-specific requirements. BILT is used in 182 countries. 

For those reasons Salute chose a long-term agreement rather than a one-year content project. Engineers carry the training with them for the length of their careers. 

5. People first: veterans, careers, and community

Salute was founded in 2013 by Lee Kirby and Jason Okroy to help veterans move from active duty into meaningful infrastructure jobs. Sanchack said the goal today is “not to find them a job, it’s to give them a career.” 

Salute does not develop or own data centers. As a lifecycle services provider, Salute supplies the expertise and people a data center needs at every stage: advising on and designing capacity, power, and cooling; commissioning new facilities by testing every system under load; operating and maintaining them day to day; and refreshing them as equipment ages and AI pushes density higher. 

That work keeps Salute’s engineers on site for the life of each facility. Their engineers  are residents of the communities where those data centers run. 

“We are the people in these cities and in these towns, and we want to make sure that it has a quality of life for all of us.” — Erich Sanchack 

Henderson described BILT’s purpose in the same terms: to enrich the lives of the people it serves. He wants every Salute technician who uses BILT to come away knowing “Erich cares about me and my family. He wants me to be safe. He wants me to be productive.” A productive technician earns more and delivers more value to the company and its customers. 

6. What infrastructure leaders can take from this

• Plan headcount and training together. A hiring target without a matching proficiency plan becomes a safety risk. 

• Treat procedures as live software, not documents. SOPs, MOPs and EOPs should update everywhere, the same day. 

• Design for sprawl. Remote AI campuses need the same standard as urban inference sites, delivered without flying everyone to a classroom. 

• Localize by default. Build regional safety and code requirements into the content, not into a separate binder. 

• Build pipelines through partners. Veteran and trade programs can supply motivated, disciplined talent at volume when training is built into the handoff. 

About Salute 

Founded in 2013, Salute is a global provider of end-to-end data center lifecycle services, from advise and design to commissioning, operations and refresh, and works with 80% of the world’s data center operators. 

About BILT 

BILT provides 3D Intelligent Instructions: interactive and XR immersive work instructions for iOS, Android and Apple Vision Pro. More than 14 million people in 182 countries use BILT, including enterprises such as Siemens and the US Air Force.