Before You Buy Another CNC: Prove the Capacity You Already Have

Before you sign for another CNC, prove your existing machines are really full: cutting metal for most of the hours you staff them, with the queue waiting in front of them rather than elsewhere. In many mid-size Indian machine shops the cell that feels full is losing hours to setups, waiting and shift edges, and a new machine starts out losing the same hours. Measure for four weeks; buy only if the numbers still say buy.

By the Mama Editorial Team. For CNC and auto-component owners facing a new OEM schedule and a supervisor who says "we need one more VMC". We don't repeat what spindle utilisation is or how it sets your machine hour rate; this is about the decision: what to measure, for how long, and what result means buy.

Key takeaways

• "We're full" is a feeling, not a measurement. A staffed machine can cut for well under half the day, and a new one inherits the same losses.
• The capex is bigger than the price tag. Indian dealer listings put a new VMC at roughly ₹18–30 lakh for the machine alone (indicative). Tooling, installation, finance and two operators at Haryana's skilled minimum of ₹18,500.81 a month come on top.
• Theory of Constraints puts buying last: exploit the constraint first; buying equipment is "a last resort".
• Run a four-week test. Log cutting, setup, waiting, no-plan and no-operator time by shift. Buy only if the constraint still cuts most of its staffed time after the cheap fixes.
• Some reasons to buy can't be measured away: missing capability (travel, 4th/5th axis, tolerance) or customer-mandated dedicated equipment.

Why the "we need another machine" feeling is usually wrong

India's shops are buying machines heavily. IMTMA's provisional estimate puts domestic machine tool consumption up about 17% to ₹31,781 crore in FY 2024-25, and ₹18,686 crore of that was imports (IMTMA Annual Report 2024-25). Much of it is justified. But on the floor, the case for one more machine usually rests on the same signals:

• A queue of WIP bins in front of the VMC cell
• Late deliveries on a part family that runs through that cell
• Operators who look busy all shift
• A planner who can't fit the next OEM schedule into the Excel loading sheet

All four fit a cell that is short of machines. They fit a cell that is short of cutting hours just as well. A machine staffed for 16 hours that cuts for 6 shows a queue, ships late and keeps its operator busy. The other 10 hours went to setup, first-off approval, waiting for blanks or the crane, an unready program, and slow starts and early wash-ups at shift edges. The loading sheet hides this because it plans cycle time × quantity as if the machine cuts all shift. A sixth machine shares the same setters, crane and material flow, and loses hours the same way.

What Theory of Constraints says about buying machines

Eliyahu Goldratt's Theory of Constraints gives an order of operations. A widely used summary of the five focusing steps runs: identify the constraint, exploit it with existing resources, subordinate everything else to it, and only then elevate it. Elevating is the step where you would "purchase additional equipment to supplement the constraint", which the same source labels "a last resort".

Applied to a machine shop:

1. Identify. Which machine or cell actually gates your shipments? It may be the heat-treatment batch, the CMM room or the one setter who proves out programs, not the cell the supervisor names.
2. Exploit. Make the constraint cut more of its existing hours: run it through lunch and shift change, stage blanks beside it, give it crane priority, check first-off at the machine.
3. Subordinate. Other machines work to keep the constraint fed, even if their own utilisation drops.
4. Elevate. Buy, add a shift or subcontract only when steps 2 and 3 can't close the gap.

Setup time is usually the biggest lever in step 2. SMED separates internal setup work, done only with the machine stopped, from external work done while it still cuts the previous job. The aim is to bring changeovers down to single-digit minutes (Lean Enterprise Institute). On a VMC: preset tools off the machine, stage the fixture kit and next blanks before the last part comes off, prove the program before the setter walks over.

The four-week capacity test

Four weeks covers a month-end rush, a slow week, both shifts and usually a breakdown. On the cell you want to expand, log every staffed machine-hour into one of five buckets:

Two checks beat the averages:

• Where does the queue sit? If WIP stacks up in front of the cell and the cell cuts most of its staffed time, it is the constraint. If the cell is waiting on blanks from the saw, the saw or stores is the constraint, and a new VMC would be one more machine waiting on blanks.
• Is "no plan" real? Unscheduled hours on a good machine while another cell is overloaded point to routing, not capacity.

A simple pass/fail rule (our working threshold, not an industry standard): treat the case for buying as proven only when, after two weeks of exploit-step fixes, the constraint is cutting for roughly two-thirds or more of its staffed time on both shifts and the queue in front of it is still growing. If it still cuts well under half its staffed hours, you have a cutting-hours problem that a new machine won't fix.

The data source decides whether the test can be trusted; end-of-shift logs round generously. For an old fleet with no controller output, the options are compared in rupees in Tracking Old CNCs Without a PLC.

Worked ₹ example: one new VMC vs the same hours from the existing cell

Take a Manesar-belt auto-component shop with five VMCs on two shifts. Planning says the cell needs about 2,000 more cutting hours a year for a new OEM part family. Figures below are illustrative. Every line marked assume is a placeholder for your own number.

The cell today - Staffed time: 16 h × 300 days (assume) = 4,800 h per machine, 24,000 h for the cell. - Measured cutting share (assume, from the four-week test): 38%, or 9,120 cutting hours. - Needed: 9,120 + 2,000 = 11,120 h, which is 46.3% of the same staffed hours.

Option A: buy a sixth VMC - Machine: B2B dealer listings for new VMCs range from about ₹18.5 lakh (TradeIndia) to ₹25–29 lakh (IndiaMART) as of October 2026. Listings are indicative, not OEM quotes; size and control change the price a lot. At the low end, with tooling, fixtures, foundation and installation, assume ₹25 lakh installed. - Finance: assume a 5-year term loan at 10%. The EMI works out to about ₹53,100/month, or ₹6.37 lakh/year. - Operators: two (one per shift) at Haryana's skilled basic minimum of ₹18,500.81/month (₹711.56/day), effective 1 April 2026 and CPI-linked (Haryana Labour Department notification No. 2/25/26-2 Lab of 9 April 2026). That's ₹4.44 lakh/year at the statutory floor. Real CNC pay is usually higher. - Power, maintenance, inserts: assume ₹1.5 lakh/year. - Hours delivered: if the new machine runs like the rest of the cell, 4,800 × 38% ≈ 1,820 cutting hours, which is short of the 2,000 needed.

Option B: recover the hours in the existing five - Spend: tool presetter, a second fixture set for the two highest-runner parts, kitting trolleys, and first-off gauging at the machine. Assume ₹4 lakh, one time. - Target: move the cell from 38% to about 46% cutting (8.3 points) out of the 62% now lost to setup, waiting and shift edges.

At an assumed ₹1,200 contribution per cutting hour, Option B's 2,000 hours are worth about ₹24 lakh/year and Option A's 1,820 about ₹21.8 lakh, so the real difference is the cost of getting them. Option B also improves Option A: a sixth machine joining a cell that cuts at 46% would deliver about 2,200 hours, not 1,820.

If the test shows the cell at 70% and still queuing, buying is the honest answer, and you have data to show your bank. Price an extra shift too: see the ₹ case for a third shift. Under the OSH Code, in force with the other labour codes since 21 November 2025, overtime is paid at twice the rate of wages, needs the worker's consent and is capped by the appropriate (usually state) government (OSH Code 2020, s. 27), so put that in the comparison.

When buying is the right call anyway

Utilisation data is not a veto. If a specific OEM enquiry is driving the question, run the RFQ capacity check on that part first. Buy without waiting for four weeks of logs if:

• The capability is missing. The part needs travel, axes, spindle power or tolerance your machines can't hold; no SMED turns a 3-axis VMC into a 5-axis.
• The customer requires dedicated equipment for a part family under its quality system. Check this against your PPAP/IATF commitments, not a guess.
• The constraint is already cutting most of its staffed time on every available shift, and the order is firm and long enough to carry the EMI.
• One machine is a single point of failure for a part that would stop an OEM line. That's redundancy, not capacity; call it that.

What cameras can and can't tell you here

A camera on the cell can show whether each machine is in cycle (stack light, door, chips, coolant), when the first chip happens, how long a changeover runs, and whether blanks and crane were there when the machine stopped. That's most of the five-bucket log, with reasons, unrounded.

It can't tell you:

• Whether a program is efficient. Air-cutting or conservative feeds look as busy as an optimised cycle.
• Capability. Tolerance, finish and Cpk come from the CMM.
• Your order book. Whether the demand is firm is a sales question.
• The cut behind a closed enclosure. The camera reads the stack light and door, not the tool.
• Anything about a particular person. Hours should be counted per machine and per shift, never as an operator ranking. If cameras cover work areas, tell workers what is recorded and why (DPDP worker notice template).

Where Mama fits

Mama reads the cameras you already have over RTSP/ONVIF, or ones we add, and every morning sends the owner a short WhatsApp note on where machine hours went yesterday: cutting, setup, waiting, or idle at the shift edges, with a rupee figure attached. It watches machines and material flow, not individuals. Weighing a new CNC? Send a short phone video of your floor and get back a camera plan for running the four-week test before the purchase order.

FAQ

How do I know if I really need another CNC machine? Measure the cell for about four weeks, splitting staffed time into cutting, setup, waiting, no-plan and no-operator time. If the constraint machine is still cutting for most of its staffed hours after the cheap fixes, and work keeps queuing in front of it, you need more capacity. If it cuts for well under half, the shortfall is in lost hours, not machines.

What does a new VMC actually cost an Indian shop per year? More than the price tag. Dealer listings run roughly ₹18–30 lakh for the machine; add tooling and installation, finance it, staff two shifts at Haryana's 2026 skilled minimum of ₹18,500.81/month, and our illustrative example reaches about ₹12 lakh a year in recurring cash cost. Use your own quotes.

Isn't it faster just to buy? Not always. A new machine needs delivery, installation and prove-out before its first saleable part. Setup and material-flow fixes can add hours from the first week.

Is a used VMC a cheaper way to add capacity? It lowers the EMI, not the lost hours. A used machine inherits the same setups, waits and shift edges, plus its own breakdown risk. Run the test first; if buying still wins, compare new and used on the hours each will really cut.