The Kennametal NSR-123B and the $2,300 Lesson That Changed How I Buy Cutting Tools

It was a Tuesday in February 2024 when Jeremy, our machining supervisor, walked into my office holding a broken insert between two fingers. "Need a replacement for this Kennametal part," he said. On the packing slip he'd taped to a sticky note was the number: NSR-123B.

I've been the administrator who handles machine shop purchasing for about five years now—roughly $400,000 a year in cutting tools and related consumables (maybe $430,000 after 2024 price adjustments, I'd have to check the system). I'm not an engineer. I'm the person who makes sure machinists and engineers get what they need without blowing the budget, breaking a purchasing contract, or creating a compliance headache. We're a mid-size contract manufacturer, around 140 employees, mostly hydraulic components for construction and agricultural equipment.

That part number started a chain of events that produced two scrapped batches, one painfully honest conversation with a Kennametal application engineer, and a permanent change in how I buy cutting tools.

The Substitute That Made Sense on Paper

In January 2024, our shop was running a stainless steel shoulder milling job that kept a CNC busy roughly 20 hours a day across two shifts. The insert that worked—the NSR-123B—had been on our bills of material for years. But the situation got complicated: our preferred tooling distributor changed their contract terms at the end of 2023, and my manager, who runs operations and has to answer to finance, was under pressure to trim tooling spend by about 8%.

When our safety stock ran out, nobody wanted to pay expedited freight to the Kennametal distributor. Sales ops—with my enthusiastic agreement—suggested we cross-reference the NSR-123B with a "comparable" insert from a cheaper vendor. Same diameter. Same basic shape. The coating even looked similar under a desk lamp. I assumed "same specifications" meant identical results across vendors. Did not verify beyond that.

In fairness, the cheaper vendor wasn't trying to deceive us. Their catalog genuinely lists a similar geometry. But similarity isn't sameness, and in cutting tools the difference shows up in edge wear and chip shape, not on paper.

What I learned is that a cutting insert part number is a spec sheet, not just a part number. Change one character and you've changed the carbide grade, the coating, or the corner radius. The NSR-123B calls out Kennametal's KC5010 grade, engineered for steel and stainless machining at moderate-to-high speeds, including interrupted cuts. The substitute had a harder grade with lower toughness. "Harder" sounds better until a machinist explains that harder carbide resists wear but is more brittle—ideal for a clean, rigid, continuous cut; terrible for an interrupted cut that hammers the edge on every revolution. (It's all published application data, by the way, in the Kennametal catalog. I just hadn't ever opened the right page.)

The first batch of 50 inserts lasted about four days before the first edge fracture. By the end of the week, tool life had dropped from around 45 parts per edge to maybe 14. Maybe 13.5—I'm mixing it up with the rework numbers. Then came the telephone call every purchaser dreads: quality found microfractures on 200 completed parts. Rework cost $2,300, including labor and machine time. All to save about $4 per insert.

The Call I Should Have Made First

After the rework, my manager asked me, reasonably, what we were going to do differently. I didn't have a good answer. So I did what I should have done before ordering anything: I called Kennametal's application hotline and asked, bluntly, what I'd missed. I was transferred to an engineer named Jack Harmon, who spent a full hour explaining cutting tool metallurgy to a purchasing administrator who couldn't tell a PVD coating from a CVD coating.

Here's what most people don't realize about Kennametal part numbers: the letters and digits aren't just inventory codes. They encode geometry, grade, and coating. The NSR-123B isn't an exotic special-order tool. It's a standard catalog item with a narrow, specific application window. The substitute looked identical. It wasn't.

Jack walked me through why KC5010 suits our stainless shoulder milling operation: the coating chemistry manages the heat and edge buildup that stainless produces, and the substrate absorbs the mechanical shock from an interrupted cut. A generic coating doesn't do that. He read the engineering recommendation straight from Kennametal's grade selector and then did something I didn't expect—he talked me out of a more expensive option.

(Should mention: Kennametal's application line has a call-back option I ignored for two days because I was convinced email would be faster. It wasn't.)

When I relayed Jack's explanation to Jeremy and Henry, our senior machinist, they just nodded. Henry had flagged the risk before the order went out. I hadn't known how to evaluate his input—he runs that machine, he's been doing it for eighteen years, and I was still new enough to trust a spec sheet over a machinist. The floor usually knows. The paperwork is still your responsibility.

About Kennametal Additive Manufacturing

The more expensive option I'd asked about: a custom tool. I'd read about Kennametal's binder-jet 3D printing of carbide—they've been expanding that capability steadily, and it keeps appearing in their product announcements. The process lets them print tool bodies and wear parts with internal coolant channels you can't create through conventional grinding. Our operation fights chip evacuation on deep bores, and I asked Jack whether we should order an additively manufactured version of the NSR-123B.

His answer was refreshingly direct: "You don't need additive tooling to fix a wrong grade. You need the right grade."

For a standard operation with standard geometry and high daily volume, additively manufactured tooling costs more per unit and takes longer to deliver. Kennametal's additive manufacturing capability is for a different set of problems: low-volume custom geometries, internal cooling features you can't grind, specialized chip-breaker designs, or jobs where cycle time matters more than tool cost. As of the 2025 catalog, at least, it's an engineering-driven solution, not a stock-item process. For a hundred identical inserts a month, conventional carbide manufacturing is faster and cheaper. For a dozen parts with an internal cooling channel that can't be ground, AM might be the only practical option.

It took me four years and an expensive scrap bin to understand that the opposite of "cheapest" isn't "most expensive." It's "correct for the application." And a supplier who tells you when you don't need their capability is a supplier you can trust when you do.

What I'd Tell Another Buyer

We ordered the Kennametal NSR-123B through our regular distributor, paid roughly 30% more per insert than the substitute, and performance snapped back to about 45 parts per edge within a week. The $4 per insert "savings" had cost us a $2,300 rework, a missed shipment, and a tense meeting with our own quality department. The spreadsheet looked much better after that.

A few things I now do differently (note to self: "check coating compatibility" is still on my office whiteboard):

• I treat the Kennametal part number as the specification. If a vendor offers something "comparable," I get an application review from Kennametal's team before switching—or an explanation of what's actually different.
• I track tool life per edge, not price per insert. The cheapest insert is only cheap if it survives the job.
• I keep grade basics in mind—ISO 513 groups hard cutting materials by application: P for steel, M for stainless, K for cast iron. Our job sits on the P/M boundary, exactly where guessing fails.
• I consider Kennametal additive manufacturing only for genuinely non-standard problems: custom geometry, chip control, internal coolant delivery. For standard stock replacement, it's not the right tool for the job.

I'm not claiming the NSR-123B is the best insert in the world. It's the right insert for our machine, our material, and our operation—and that's a different, load-bearing statement. If you're thinking of cross-referencing it with a cheaper substitute, get the application review first. I do not trust "comparable" anymore. That trust cost me $2,300 to lose. Next time someone asks me whether Kennametal is expensive, I'll tell them this story.