# Who’s actually building AI in hardlines retail, and who’s just talking.

Canonical URL: https://sgabenchmarking.com/ai-org-diagnostic/

The AI Org Diagnostic scores what each company staffs, ships, files and says across Home Depot, Lowe's, Tractor Supply, Ace Hardware, Floor & Decor and Menards , measured the same way for every company from public data. The same diagnostic run on a different sector: [AI org benchmark for US equipment rental](https://sgabenchmarking.com/ai-org-diagnostic/equipment-rental/), across eight rental companies.

## Executive summary.

- Home Depot – a 446-person AI and data bench, 51 AI patents, and shipped tools it barely talks about.
- Lowe’s – 333 measured roles (40% offshore, almost all in India) and AI mentions that tripled across five quarters.
- Tractor Supply – a 25-person bench with machine learning across 90% of replenishment forecasts.
- Ace Hardware – Hey ARMA in 2,300+ stores with roughly one to two core AI staff in-house.
- Floor & Decor – an 8-person bench, two open AI roles, zero AI mentions in five quarters of earnings calls.
- Menards – zero AI-titled people, zero postings, zero disclosures, zero press.

## Home Depot and Lowe’s field real AI and data benches. The rest of the cohort fits in one conference room.

Pound for pound, Lowe’s leads: 12.1 combined roles per 10,000 employees.

| Company |  | AI + data-science headcount 1 | Employees | Per 10k employees |
| --- | --- | --- | --- | --- |
| Home Depot |  | 446 | 470,000 | 9.5 |
| Lowe’s |  | 333 2 | 276,000 | 12.1 |
| Tractor Supply |  | 25 | 52,000 | 4.8 |
| Ace Hardware 3 |  | 10 | 13,000 | 7.7 |
| Floor & Decor |  | 8 | 13,723 | 5.8 |
| Menards |  | 0 | 45,000 | 0.0 |

## Lowe’s talks the most, and now has the bench to back it. Home Depot builds more than it says.

Lowe’s per-call AI mentions roughly tripled across the five quarters , and unlike most loud talkers, its bench is real. Floor & Decor builds without saying a word.

## How the AI org is layered.

Home Depot and Lowe’s run full four-layer organizations : exec sponsors, a director tier, a deep IC base. Ace’s ten people include no one above manager level.

| Company |  | Mgr+ / IC 1 | Exec+ / Mgr 1 |
| --- | --- | --- | --- |
| Home Depot In-house at scale + Google Cloud |  | 0.67 | 0.21 |
| Lowe’s Partner-branded: OpenAI + NVIDIA |  | 0.58 | 0.37 |
| Tractor Supply Partner-led + small platform team |  | 0.47 | 0.60 |
| Floor & Decor Early in-house pilot |  | 1.00 | 0.33 |
| Ace Hardware Vendor-built; builder undisclosed |  | 0.25 | 0.00 |
| Menards No visible AI activity |  | — | — |

## Where the AI org sits.

Home Depot’s org is ~95% US-based. Lowe’s runs the only offshore-heavy model in the cohort: ~39% of its AI org sits in India (Lowe’s Services India, Bangalore).

| Company |  | Offshore share 1 |
| --- | --- | --- |
| Home Depot |  | 5% |
| Lowe’s |  | 40% |
| Tractor Supply |  | 4% |
| Floor & Decor |  | 13% |
| Ace Hardware |  | 0% |
| Menards |  | — |

## How these numbers are built.

### Measure the bench

Combined AI + data-science headcount from public workforce data, resolved to each corporate family before counting: dedicated core-AI/ML and data-science titles only, the same way for all six companies.

### Read the flow and the narrative

Open roles from public job postings screened to dedicated AI and data-science titles, core-AI hires over the trailing 12 months from position-start dates, and every AI mention on earnings calls 2025Q1–2026Q1, read against the measured bench.

### Count only what we can see

Counts are floors of dedicated AI- and data-titled roles, not a census. Disclosed products, patents and performance claims are recorded as disclosed, never as verified. The same [outside-in discipline](https://sgabenchmarking.com/index.html#how) as our G&A benchmarks.

## Four of six have disclosed customer- or associate-facing AI. Only Lowe’s discloses performance numbers.

Everything below is what the company (or its vendor) has said publicly; nothing here is independently verified, and no cohort company has disclosed dollar-denominated AI ROI. Floor & Decor and Menards have disclosed nothing.

- Magic Apron gen-AI shopping assistant (Mar 2025)
- Agentic AI tools w/ Google Cloud (NRF Jan 2026)
- AI materials-list builder for Pros (beta Nov 2025, national 2026)
- AI voice agents in stores (pilot); Gemini Enterprise for associates
- Mylow AI shopping advisor w/ OpenAI (Mar 2025)
- Mylow Companion for all associates, 1,700+ stores (May 2025)
- AI restock prioritization from real-time out-of-stock data
- NVIDIA digital twins / Omniverse
- Hey GURA gen-AI wearable assistant for every team member
- ML across 90% of replenishment forecasts (2025Q2 call)
- AI photo-assessment of store resets
- Delivery-hub network scaling w/ AI; expanded OpenAI relationship
- Hey ARMA AI assistant on the ARMA handheld, 2,300+ stores (Feb 2026)

## Home Depot is still hiring AI and data talent. Menards shows nothing, again.

Open AI- and data-titled roles plus core-AI hires over the trailing 12 months. Who’s hiring matches who’s built. Tractor Supply’s stock is built, flow paused; Ace’s zero is consistent with a vendor-built model.

| Company |  | Open AI + data-science roles 1 | Core-AI hires, trailing 12 mo 2 |
| --- | --- | --- | --- |
| Home Depot |  | 26 | 20 |
| Lowe’s |  | 3 | 3 |
| Floor & Decor |  | 2 | 1 |
| Tractor Supply |  | 0 | 1 |
| Ace Hardware |  | 0 | 1 |
| Menards |  | 0 | 0 |

## What this example can, and can’t, tell you.

### Free: this example

- Six named companies, measured with one identical method
- The combined AI + data-science bench, with the core-AI split and density per 10,000 employees
- Hiring flow and earnings-call narrative read against the measured bench, not taken at face value
- Disclosed use cases, performance claims and AI patents per company, labeled as disclosed, not verified

### The full diagnostic: what we deliberately withhold

- The named-profile roster behind every count
- Your company and your chosen peer set, measured identically
- Hiring flow and narrative tracked quarterly as a trend, not a snapshot
- Full disclosed-use-case and patent detail per peer
- A walkthrough read-out call

## FAQs

### What counts as an “AI role,” and why combine AI with data science? +

### Why are these counts a floor, not a census? +

### Where does the data come from? +

### How is Ace Hardware’s number comparable when it’s a co-op? +

### Does a big bench mean deployed AI, or verified ROI? +