With inventory distortion now behind us, we raise our utilisation assumptions for CSM. Still, two concerns remain: 1) recovery of orders from IDMs will take time, since internal capacity is still available; and 2) despite its rights issue, CSM needs to raise more funds. We remain NEUTRAL as we await clarity on funding.

Mounting design costs for finer process nodes have seen some IDMs discontinue in-house R&D in favour of alliances. The IBM-CSM alliance is seeing increased acceptance, having proven its platform as a viable alternative to TSMC. AMD, Freescale, Infineon, ST Micro, and now Toshiba are adopting the IBM platform.

High transition costs to 65nm have effectively delayed the 65nm ramp at CSM’s new wireless customers. Limited transactions with these customers on older nodes and continued demand woes exert downward pressure at CSM.

We are seeing a recovery in orders from the fabless chipmakers as the channel inventory correction comes to an end. Exhibit 1 shows the sales and inventory of the top 30 fabless customers, which together account for about 70% of Asian foundry demand. Our forecast is for 1Q09F sales to fall US$11bn — off 70% from the peak of US$17bn in 3Q08. We believe this decline is distorted by inventory correction in the channel. On our estimates, end-product demand fell 9% q-q in 4Q08 and 20% q-q in 1Q09 — with the difference (vs the 70% sales decline) driven by channel inventories being worked down. The dotted line shows our reconstruction of sales, had inventory correction not had an impact. We expect normalisation of channel inventory to be complete by end-1Q09F. Intel’s comment in its results call yesterday that channel inventory in the PC segment was back to normal levels reinforces our view that inventory correction has run its course.

For 2Q09F, we see global utilisations recovering to 65-70%, from 45-50% in 1Q09, as inventory distortion washes out. However, we think order growth at foundries is likely to come only from fabless makers, since IDMs first have to fill in-house fabs. For 2H09F, we see further improvement in foundry utilisations being driven by: 1) seasonal demand growth in 2H09F; and 2) partial recovery of orders from IDMs. We forecast CSM’s utilisations at 54%, 60.4% and 59.8% in 2Q09F, 3Q09F and 4Q09F (earlier: 39%, 51% and 49%, respectively).

Given the better outlook for utilisations, we reduce our net loss forecasts — for FY09F, we now forecast a loss of US$299mn ex one-offs (previously: loss of US$336mn); for FY10F, we bring down our net loss expectation to US$172mn (ex one-offs), from US$203mn earlier. Upside risks to our expectations: 1) faster-than-expected reduction in breakeven utilisations at CSM (note: CSM currently expects a reduction in breakeven utilisation to 75% by end-4Q09F, from the mid-80s in 4Q08); and 2) faster-than-expected 65nm ramp.

We maintain our NEUTRAL view of CSM, since we remain concerned about the need to raise further capital in FY10F. On the new equity base post completion of the rights issue, we set a price target of S$0.14 (method unchanged; earlier: S$0.22, before adjusting for rights issue). We derive our price target by applying our FY09F BV forecast to the average forward P/BV of CSM’s closest peers, UMC and SMIC.