In recent months, several friends of mine have asked if artificial intelligence will replace the work that we do as supply chain network engineers.  In fact, as I type, artificial intelligence offers suggestions on how I want to end a sentence as I am typing it.  That simple example suggests the benefits and drawbacks of artificial intelligence.

For the sake of this article, let’s start with the following assumption and question for Chat GPT: “A company has a demand footprint that mirrors the US population, imports its product from Asia via container, and distributes out of Reno, NV.  Where should my next warehouse be located?”  Let’s see what Chat GPT recommends…

Chat GPT’s Response:

Current Setup Summary:

  • Demand Distribution: Mirrors the U.S. population (i.e., highest demand on the East Coast, significant in Midwest, less in Mountain West and West).
  • Inbound Logistics: Imports product from Asia via ocean container.
  • Current Distribution Hub: Reno, NV (serves as a West Coast hub).
  • Likely Goal: Improve delivery times, reduce shipping costs, and balance East/Midwest distribution.

Key Constraints and Goals:

  • Reduce time and cost to serve East and Central U.S.
  • Maintain efficient inbound logistics from Asia.
  • Leverage intermodal or port access (optional, if importing directly to the new location).
  • Optimal Next Warehouse Locations (Shortlist)

Here are top contenders, based on logistics, population center proximity, and cost efficiency:

  1. Columbus, Ohio
  • Near the U.S. population center.
  • Great highway access (I-70, I-71, I-75), rail, and intermodal terminals.
  • Within 1-day truck reach of ~50%+ of U.S. population.
  • Moderate labor and real estate costs.
  • Best for: East Coast and Midwest coverage, fast fulfillment.
  1. Kansas City, MO / KS
  • Central location allows bridging West (Reno) and East efficiently.
  • Cheaper than Eastern locations.
  • Best for: Centralized balancing between coasts, lower cost hub.
  1. Chicago, IL (or outskirts like Joliet, IL)
  • Major logistics and transportation hub.
  • Cons: Higher costs, congestion.
  • Best for: High-volume Midwest hub.
  1. Atlanta, GA
  • Southeastern U.S. population center.
  • Strong distribution infrastructure (airport, rail, highway).
  • Best for: Southeastern U.S. and East Coast penetration.

Recommendation:

Build your next warehouse in or near Columbus, Ohio.

Reasons:

  • Strategically close to national population center.
  • Complements Reno by extending reach eastward.
  • Strong intermodal connectivity and labor availability.
  • Cost-effective compared to coastal or urban mega-hubs.
  • Reduces shipping times and costs for 50–70% of demand footprint.

Let me know if you’d like a map-based analysis, cost comparison, or transport time modeling for a deeper decision.

END OF CHAT GPT

Having asked a relatively straightforward question, Chat GPT gave a pretty good answer.  Where Chat GPT falls noticeably short of replacing the work we do is in the following areas.

First, understanding additional details, such as transportation modes, service requirements, top customer considerations, when trucks hit capacity based on space or weight, etc.  The list of additional considerations is often 20 or more.  With that said, the locations it suggests and the reasons it suggested them are still pretty good.

Second, being able to understand the value of a second location, so I asked, “What would that be worth to our organization?”

Chat GPT’s response:

Assumptions (Customizable)

  • Metric Value
  • Annual shipments 500,000 orders
  • Average order weight/size Small parcel (5–10 lbs)
  • % of demand East of Mississippi 65%
  • Current avg. East Coast delivery cost $10/order
  • Current avg. delivery time (East) 4–6 days
  • Warehouse cost (Columbus, OH) $3M/year (lease, labor, ops)
  • Outbound savings per East order with OH hub $3–$5

Potential Savings

  1. Outbound Shipping Cost Savings
    1. By serving Eastern U.S. orders from Ohio instead of Nevada, you’re reducing zone distances (Zone 7+ down to Zone 3–4), which can lower parcel shipping or LTL rates.
    2. Orders shifted to Columbus warehouse: 65% of 500,000 = 325,000 orders/year
      1. Savings per order: ~$4 (zone reduction)
    3. Annual Shipping Savings:
      1. 325,000 orders×$4=$1.3M/year

Delivery Time Improvement → Customer Retention + Revenue Impact

  1. If you reduce delivery time from 5 days → 2 days for Eastern customers:
    • Higher customer satisfaction
    • Lower cart abandonment
    • Potentially higher repeat purchase rate or conversion
  2. Conservative Estimate:
    • 5% lift in sales to Eastern customers
    • Avg order value = $100
    • 325,000 orders → 16,250 more orders
  3. Revenue Impact:
    • 16,250×$100=$1.625M/year

(NOTE: This is revenue, not margin.)

Inventory Positioning / Working Capital Optimization

  1. Faster delivery from a distributed network allows leaner inventory per site and better regional availability.
    1. Estimated Inventory Efficiency Gain: 5% reduction in safety stock needed
      1. Let’s say your average inventory value is $20M
      2. 5% = $1M freed up working capital (one-time)

Risk Reduction / Business Continuity

  1. Diversifying away from a single-warehouse model reduces disruption risk (fires, weather, labor issues, etc.)
    • Hard to quantify, but insurers may offer lower premiums; also lowers existential fulfillment risk.

Total Estimated Annual Value

  • Shipping cost savings $1.3M/year
  • Revenue lift from faster shipping $1.6M/year
  • Inventory carrying cost savings $50K–$100K/year
  • Risk mitigation (qualitative) Medium–High
  • Total Estimated Annual Benefit $2.9M–$3.1M
  • Offset Against Warehouse Costs
    • Assuming:
      • Facility, labor, and ops = $3M/year
    • Net Impact: Near breakeven in year 1, but:
    • Payoff improves if sales grow
    • Volume consolidation lowers cost per unit
    • Strategic advantage grows with network effects

Bottom Line

A Columbus warehouse could generate $2.9M+ in direct annual value, not including intangible strategic gains.

If you can confirm shipment volume, current costs, and customer density, I can fine-tune the ROI model further. Would you like a version in Excel or Google Sheets?

END OF CHAT GPT

What should we conclude from this.

First, Chat GPT and other artificial intelligence platforms can do a pretty good job of suggesting where you should consider additional locations.  However, most logistics professionals can generate that same list of locations and reasons for consideration simply by looking at a map.  It leaves me with the question, “Did Chat GPT provide any value with the first question?”  Not really.

Once the question went beyond simply locations to consider, Chat GPT needed to define assumptions to start to answer what the selected city would be worth.  This is where the current artificial intelligence algorithms fall short.  It appears that the answer to this question did a pretty good job of writing a request for proposal, not answering the question.  In fact, the last sentence before the communication ends is where our network optimization engineering teams begin.

The first step in our consultation is data collection, including the following data sets:

  1. Item master
  2. Customer Master
  3. Facility Master
  4. Transactions
  5. Shipment
  6. Inventory
  7. Growth

In conclusion, someday artificial intelligence may be able to replace the work a network optimization engineer does.  The AI will require unfettered access to the data listed above and for the foreseeable future will require guardrails to ensure suggestions remain within the bounds of your current business.  I welcome the AI industry to our field.  It highlights the risk of overconfidence in the face of limited information.
 
—Jeff Schmidtke, St. Onge Company
 
 

Subscribe to our Blog!