Semiconductor Supply Chain
Foundry capacity allocation is the process of deciding which customers' designs actually get wafer starts when total demand exceeds available capacity — and during a genuine shortage, wanting wafers isn't the deciding factor at all. The foundry is weighing qualified design readiness, forecast credibility, contractual commitments, and overall customer relationship value, because allocating a scarce wafer slot to a design that isn't actually ready to consume it productively wastes capacity everyone downstream is desperate for.
Forecast quality is the single input that actually lets a foundry commit scarce capacity in advance, and weak forecasts carry a real, visible cost to whoever submits them. A foundry has to plan wafer starts months ahead of actual production, which means it's effectively betting fab time on a customer's stated future demand. A customer with a track record of accurate, reliable forecasting gets allocation priority precisely because the foundry can plan around their numbers with confidence; a customer whose forecasts have historically run wildly optimistic or consistently missed either loses priority in the next allocation round or, if accommodated anyway, risks leaving the foundry holding excess, unsellable wafer inventory that the forecast promised would be needed.
Design and node readiness measures whether a customer's chip can actually consume the wafer capacity being requested, which matters just as much as wanting the capacity in the first place. A design that's fully qualified against the target process node, with finalized masks and a validated test plan, can start production the moment a wafer slot opens and begin generating usable die immediately. A design that's still mid-qualification, or that needs process tweaks once early silicon comes back, risks sitting on allocated capacity without actually using it productively — exactly the outcome a foundry tries hardest to avoid when every wafer slot is this valuable. This is why node and package readiness functions almost as a gate on allocation itself, not just a nice-to-have.
Customer tier reflects accumulated relationship value — historical volume, contract terms, and strategic importance — and it's a real, if sometimes uncomfortable, factor in how allocation decisions get made. A customer who has reliably purchased large volumes over many years, honored forecasts, and maintained a close technical relationship with the foundry's own engineering teams generally receives priority over a newer or smaller customer making the same capacity request during a shortage, independent of either customer's design quality. Foundries justify this by pointing to the long-term trust and lower execution risk that an established relationship represents, but it also means that during genuine shortages, allocation outcomes aren't purely a meritocracy based on design quality or payment willingness alone.
Supply-chain completeness around the wafer itself — substrates, packaging capacity, and test capacity — matters just as much as the wafer slot, because winning the wafer allocation accomplishes nothing if the next step in the chain can't absorb the output. A customer who secures wafer capacity but hasn't also lined up substrate supply, OSAT packaging capacity, and test slots simply shifts the bottleneck one step downstream instead of resolving it — the wafers get fabricated but then sit, unable to become finished product. This is exactly why sophisticated allocation planning treats the wafer as one link in a longer chain rather than the sole scarce resource, and why foundries increasingly ask customers to demonstrate readiness across that entire downstream chain, not just at the fab gate, before committing scarce wafer capacity to them.
Allocation decisions made during a shortage have consequences that persist well past the shortage itself, which is exactly why they carry so much long-term strategic weight. A customer deprioritized during a capacity crunch may permanently diversify toward a competing foundry rather than return once capacity loosens up, and a foundry that mishandles allocation during a critical shortage can damage relationships with major customers for years afterward. This is why the biggest, most sophisticated chip customers increasingly pursue multi-foundry sourcing strategies and long-term capacity reservation agreements — hedging against exactly this kind of allocation risk before the next shortage arrives, rather than hoping to win a favorable allocation decision after one has already begun.