A warranty is the precived average Lifespan of the product based on manufacturers testing, after warranty expires i would replace the PSU
Not quite. if the warranty length was the
average lifespan, then statistically, half of all PSUs would fail under warranty and the company would rapidly go out of business.
No, warranties are a gamble of how long companies think their units will last without failure. If they think 95% of products will last the warranty period, then they have to write off 5% of all units made as a loss, incurring additional costs through return shipping and RMA processing.
The calculation varies from product to product, and from company to company, but a PSU model is likely an abject, total failure if more than 4-5% of the units die before the warranty runs out. Manufacturer margins aren't that big because the PSU market is a long-established one with a huge amount of competition. If warranty returns are even 6-7% then they've suffered a huge setback and will need to recover from that using funding from elsewhere.
IMO if PSU has a 5 year warranty, 98% of those PSUs will last 5 years, even those that are run 24/7, and it's not unreasonable to expect them to last 10-15 years. I see
beige PCs occasionally that have been running in a machine shop or lab since they were built in the '90s and they're still going after 30 years with the original PSU. Remember, warranties are a marketing department thing and they obviously favour company profit over consumer rights, meaning they are pessimistically short by a wide (and profitable) margin.
@jonnyGURU,
@crmaris,
@TheLostSwede are likely far better positioned to comment on PSU longevity and BOM costs/failure rates.
To be clear I don't see you want to save here, but just use such hitting hard example for anybody reading it 
