Should we commit $400M more to memory chip inventory?

Memory chips are being discontinued. Prices are up 10×. No dip is coming, and you still ship products that need them.

It's the memory inside the things you ship, and it's disappearing.

These memory chips aren't the newest, that's the problem. Chipmakers are abandoning them to build pricier memory for AI. The supply you depend on is being redirected to the data centers, and what's left is getting scarce and expensive fast.

Routers & switchesIndustrial controllersVehiclesMedical devicesServers & storage
≈10×

Memory chip spot price since early 2025, a part that used to be an afterthought

~20%

of 2024's memory chip output still being made, the rest of the capacity went to AI memory

2028

earliest that analysts expect any meaningful relief on supply

Waiting is dangerous. So is overcommitting.

Wait, and supply vanishes, there's no dip coming. Overbuy, and a softening market leaves you holding millions in a part your own roadmap is retiring. The real call is never "buy" or "wait", it's balancing supply, demand, risk and working capital at once.

The price chart can't answer this. Your business can.

Price shock
Memory ↑10×

Procurement

Commit cash now, or gamble on waiting

Inventory

Over-buy ages into write-offs; under-buy runs dry

Production

No chips, no product, the line stops

Commitments

Orders you've already promised customers

Cash & Margin

Where every move finally nets out

Change one, and you change all five. That's a decision, not a calculation.

The decision, inside the Ergodic console.

One state of the world, every move scored against it, and the causal trace behind each number, live in the product. Every figure traces back through the causal graph, node by node.

ergodic.ai / console — demand & inventory
Ergodic
Decisions
Entity graph
Root cause
Forecasts
Scenarios
Wait & see
Buy everything
Secure & hedge
Step 03 · Explore

The $400M call, and what it costs

Secure & hedge · recommended
CASH COMMITTED
$387M
−$128M vs all-in
capital outlay
STOCKOUT RISK
5.8%
−32.7 pts
unmet demand
WRITE-OFF
$20M
−$92M vs all-in
obsolete stock
COMMITMENTS
98%
demand served
top accounts first
Inventory cover · weeks
secure & hedge recovers cover; wait & see drains
Demand vs supply · indexed
supply exits faster than demand falls
Why · drivers & causal trace
Memory chip supply exit → narrowing buy window
Demand uncertainty → phased hedge
Margin priority → protect top accounts

Every move, side by side.

Scenario
Cash
Stockout
Write-off
Served
Wait & see
$255M
38.5%
$0M
61%
Buy everything
$515M
0%
$112M
100%
Ergodic's pick
$387M
5.8%
$20M
98%

Secure enough now, hedge the rest. Supply is exiting, so waiting is a trap. But demand is uncertain, so buying everything strands capital in a part your own roadmap is retiring. Ergodic secures ~96% of a year's need now, phases a small slice as a demand hedge, and protects your highest-margin customers first.

Against buy-everything: −$128M cash and −$92M write-off, for only ~6 points less service. Against wait-and-see: it removes a ~$323M line-down exposure and a 38% stockout risk. Every figure traces back through the causal graph, node by node, no black box.

A flight simulator for your biggest decisions.

Pose a decision and watch Ergodic work it: grounding in reality, simulating demand, cascading the consequences, then ranking the moves.

Analyse a decision
01 · Contextualise

Grounds the decision in what's true: the memory chip supply exit and price spike, your real demand history, the orders you've promised, and the cash on hand.

Run your next big call in simulation first.

Bring us the decision that keeps you up at night and watch it play out before you commit.

Talk To An Expert