Last month we quoted a BESS frame project for a European integrator — 2400 aluminum frame sets, delivered over 6 months. The RFQ came with 47 pages of technical requirements, including flatness specs that would make an aerospace engineer nod approvingly. Three other shops turned it down because they couldn't hold 0.05mm flatness across a 1200mm span at that volume.

We took it. Not because we're special — because we've been building toward this exact type of work for three years.

The global BESS (Battery Energy Storage System) market is projected to hit $120 billion by 2030. That's not a forecast from a consultancy trying to sell you a report — that's the combined capacity of announced projects from actual utilities and grid operators. China alone installed 48GWh of new storage capacity in 2025.

Every single one of those installations needs CNC machined parts. Frames, cooling plates, busbar links, terminal blocks, battery clamps, mounting brackets, EMI shields. A typical 100kWh commercial BESS cabinet contains 35-50 precision machined metal components. A utility-scale containerized system? 200+.

Do the math. 48GWh of new capacity in China alone, each GWh needing roughly 10,000 cabinets, each cabinet needing 35-50 machined parts. That's 16.8 to 24 million precision machined components per year. Just from China installations. The global number is 3-4x that.

Most CNC job shops are still quoting pump bodies and general industrial brackets. The ones who've noticed the energy storage wave are either drowning in orders or scrambling to retool.

BESS components sit at the intersection of three disciplines that don't usually talk to each other: structural engineering, electrical engineering, and thermal management.

A battery frame isn't just a structural box. It's the mounting platform for cells that expand and contract 2-3mm over charge cycles. It's the alignment reference for busbar connections carrying 400A. It's the structural support for cooling plates that have to maintain <0.1mm flatness to ensure uniform thermal contact.

Miss any one of these and you don't get a callback — you get a field failure.

The tolerance requirements aren't aerospace-tight, but they're consistent-tight. +/-0.01mm on a pump body is impressive. Holding +/-0.01mm across 2000 identical parts is a different conversation entirely. BESS customers buy in volume — 500, 2000, 10000 units — and they expect every piece to be the same as the first article you qualified.

Then there's the material expertise. BESS runs on copper (C11000, C10200) for current-carrying parts, aluminum (6061, 5052) for structure and cooling, and stainless steel (304, 316) for clamping and mounting. If your shop only knows steel and aluminum, you're leaving a third of the bill of materials on the table.

Nobody talks about this, but machining copper at volume is a completely different skill set from machining steel or aluminum. Copper is gummy. It work-hardens. Chips don't break — they string. Taps break in deep holes because the material won't let go.

We went through 18 months of process development before we could reliably tap M10 holes 25mm deep in C11000 at production volumes. That's not an exaggeration — 18 months of tooling trials, parameter adjustments, and a lot of broken taps before we found the combination of rake angle, coating, peck cycle, and lubrication that gives us 300+ holes per tap.

Most shops bidding on BESS copper work haven't done that homework. They'll quote it based on their steel experience, hit problems on day one of production, and then the customer is waiting while the shop figures out that you can't machine copper like you machine 4140.

Liquid cooling plates are the highest-precision component in a BESS system, and the one most likely to be quoted wrong.

The spec looks simple: machine channels in a flat plate. But the flatness requirement — typically <0.1mm across the full surface — means your entire process has to be built around that single metric. Material stress, machining sequence, fixturing method, even the order of operations matters. Machine the channels first and then drill the inlet/outlet ports? Better make sure the drilling doesn't introduce stress that warps the plate. Machine from both sides? Better have a fixture that doesn't introduce location error when you flip the part.

We've seen plates from suppliers that looked fine on the CMM but failed in assembly because the flatness was measured at room temperature but the installation happened in a 40°C desert installation site. Thermal expansion of a 600mm aluminum plate is 0.4mm over a 20°C range. If your flatness budget is 0.1mm, that's game over.

The fix isn't tighter machining tolerances — it's designing the system to accommodate thermal movement. But that requires a shop that understands why the tolerance matters, not just what the number on the drawing says.

If you're a BESS integrator sourcing machined components, here's what actually matters — beyond the ISO certificate on the wall:

Volume consistency proof. Ask for SPC data from a recent production run. Not first article inspection — that proves you can make one good part. SPC data proves you can make a thousand.

Material-specific expertise. If they're going to machine copper busbar links, ask how many copper parts they ran last month. If the answer is "we machine all materials," that means copper isn't a core competency.

Surface finish on current-carrying surfaces. A rough contact surface means higher resistance. Higher resistance means heat. Heat in a battery system is the enemy. Specify Ra 0.8 or better and verify they actually measure it.

Plating management. Tin plating on copper isn't complicated, but the thickness matters. Too thin and the copper oxidizes through the tin in 2 years. Too thick and you waste money. Ask if they verify plating thickness with XRF or just trust the plating vendor.

Leak testing on cooling plates. If they're machining cooling plates, ask about their leak testing protocol. "We can do that" isn't the same as "we test every single plate to 1.5x operating pressure."

The BESS market is going to need machined parts in quantities that dwarf anything the CNC industry has seen since the smartphone boom. The shops that will win are the ones investing in process development now, not the ones waiting for orders to show up.