In short
Tape-outs usually slip because a change in one team is discovered late by the teams that depend on it, not because the engineering failed. About 75% of chip projects finish behind their original schedule (Siemens / Wilson Research Group, 2024). The fix is not a better plan; it is shortening the time between a change happening and its owner knowing.
How common are tape-out slips?
Very. The 2024 Wilson Research Group study, commissioned by Siemens, found that about 75% of chip projects were behind their original schedule, and that only 14% achieved first-silicon success — the lowest figure in the twenty years the study has run.
What does a slip look like from the inside?
Take a real pattern. One hundred and four days before samples, verification flags that coverage on one block is off its trend. Eighty-seven days out, two more engineers are requested and the request is deferred to a quarterly review. Sixty-two days out, physical design flags a timing impact. Forty-one days out, the block's freeze is deferred by two weeks; the cascade is predicted but not escalated. Twenty-eight days out, an integration regression fails and an ECO loop begins. Samples miss by seventeen days.
Asked afterwards, the team says verification was late. The actual root cause was a capacity request deferred eighty-seven days out. Every signal was visible on the day it occurred.
What are the usual causes?
Four patterns account for most slips:
- A coverage slope that no longer reaches the date. The curve flattens weeks before anyone calls it late.
- A late engineering change in physical design, whose effect on timing, power and area is discovered downstream.
- A third-party IP delivery that moves, announced in an email nobody connected to RTL freeze.
- A foundry shuttle or capacity change that shifts the whole back end of the schedule.
Why doesn't the weekly review catch it?
Because a status report is what people remembered to type, a week ago. It is self-reported, assembled by hand, and already out of date when it is presented. The typical gap between a slip and the leadership team knowing is one to two weeks — by which time the cheap fixes are gone.
How do you catch a slip earlier?
Read the evidence rather than the status. A milestone is not done because someone said so; it is done when the regression passes, the timing report closes, or the vendor's delivery lands. Tymeline holds each commitment against that evidence and raises the alarm the hour it diverges, with the downstream chain attached: which blocks absorb the delay, which one doesn't, and which date is now at risk.