Initiated Novellus NVLS Sell
Micron is the first DRAM maker that develops copper process for DRAM production, with fellow technology leaders also striving to advance their production in this field.
Not only DRAM but also Micron's NAND is based on copper.
Most likely NVLS is going to benefit the most compared to its peers
DRAM and probably also NAND are retooling for copper - that now is a fact;
both NVLS and LRCX are tool of record at Micron as far as I understand for copper interconnect: NVLS for deposition (copper plating) and Lam for oxide etch.
NVLS specifically targeted CVD for memory makers in conjunction with copper providing high throughput as well as low cost of ownership.
Micron is transfering copper process to Nanya and Inotera (Nanya / Qimonda JV) will probably gobbled up by Micron.
Again this is not just a shrink - that's major change in the backend process for memory guys.
LRCX should be very strong at Hynix and Samsung as well
Micron is the first DRAM maker that develops copper process for DRAM production, with fellow technology leaders also striving to advance their production in this field.
DRAM: The 50 nm challenge for the DRAM industry; NAND Flash: 2H08 memory card shipment to increase almost 30% HoH; PC: Niche market segment has derived from low-cost notebook
Published Jul.8, 2008
The 50 nm challenge for the DRAM industry
Advancement of semiconductor production encounters physical barriers/limits at every era. Despite industry players always worry about whether a breakthrough for a new process is achievable, their worries vanish via technology advancement or innovation.
In the DRAM industry, traditional ArF dry lithography equipment is still the mainstream fabrication tool, with its physical limit being identified at 65nm. This implies that for any design geometry that is smaller than 65nm, shift to immersion tool is necessary. (Note: Immersion tool utilizes water to replace air as media in between lithography lens and wafer surface for circuitry patterning. Since water delivers a higher refraction rate, smaller linewidth resulted thus breaks physical limit.) Vast capex required for process node advancement, especially procurement of expensive immersion tool, is a big challenge for those DRAM makers who are still bleeding in red.
Process node shrinking is the most efficient way to lower manufacturing cost. Most DRAM technology leaders had their mainstream technology migrated to 70nm. As traditional ArF dry tools have their physical limit beyond 65nm, they have no alternative but to opt for expensive immersion tool if they wish to keep their foothold in the market. When it comes to cost issue, it will be those deep-pocketed to take the leap first. In another words, a solid financial background will judge respective DRAM makers' technology competitive edge.
ASML, Canon and Nikon are the current three major immersion tool suppliers. The number-one vendor ASML has shipped over 70 sets of immersion tools so far, mainly for logic IC makers. Key immersion tools from ASML include Twinscan XT1700i and XT1900Gi. Of which, XT1900Gi could support process node down to <40nm and shipments to more than 10 customers have begun since 3Q07. Nikon starts supplying its NSR-S609B for 55nm and 45nm process nodes from 2006. It starts shipping NSR-S610C for 45nm process from 2007.
Besides advancing process node, DRAM technology breakthrough could be done via introduction of copper process. In contrast to the present aluminum process, copper process is 40% faster, 30% lower in cost and 30% thinner in linewidth.
Natural feature difference between memory and logic IC means a complete technology transference is impossible. Despite the fact that copper process delivers the aforementioned advantages, as well as a better electric conductivity, the contaminating issue comes with better conductivity also need higher skill to take care of..
Micron is the first DRAM maker that develops copper process for DRAM production, with fellow technology leaders also striving to advance their production in this field.
Handset and DSC entering high season; 2H08 memory card shipment to increase almost 30% HoH
1H08 contract price for 8Gb and 16Gb NAND flash chips dropped 37.5% and 39%, respectively. Excess supply was the main reason. Demand for 1H08 was bleak because handsets and DSCs had been experiencing seasonal slowdowns, with shipment down 4.4% and 6.1% HoH. The memory card used in these applications account for more than 40% of NAND Flash consumption. At the same time, bit supply increased almost 50%, as suppliers began to master the 5xnm process and build more 12-inch capacity. As a result, both chip and card price steadily trended down.
Whether price can reverse the trend will be the main topic for the rest of the year. Branded card suppliers had kept inventory within a week in order to prevent profitability erode from falling inventory value. For the rest of the year, the sell-through during September's back-to-school sales, November's Thanksgiving and December's Christmas will be the key to profitability.
DRAMeXchange forecasts handset shipment can increase 10% HoH to 657mn sets in 2H08. DSC shipment began to pick up in 2Q08, and we forecast DSC shipment to increase 26% HoH to 75.9mn sets, as shown in Chart 2. On top of these, PDA, GPS, digital video and game console will drive demand. We estimate 2H08 demand for memory card will increase 28% HoH to almost 100mn pieces.
While the increase in demand is foreseeable, suppliers' capacity allocation will still have a key impact on NAND flash chip prices. If the upstream suppliers manage well their shipments, there should be a strong urge for inventory build-up given the lean retail inventory. As a result, price should go up, but the duration will depend on the upstream's supply strategy.
NAND Flash contract price review in 1H July
NAND Flash average contract price roughly stayed flat in 1H July, due to weaker demand in NAND Flash market. As the memory card application, which are a major source of NAND Flash consumption remained weak, the news of Apple's 3G iPhone launch in mid-July could not stimulate the NAND Flash market. The NAND Flash price is likely to stabilize in the short term, due to new demand from smart phones & low-cost PCs in 3Q08. We expect the contract price to gradually rebound, as the traditional hot season demand slowly recovers after mid-3Q08.
Niche market segment has derived from low-cost notebook
Business potential from low-cost notebook has long been questioned despite respective vendors are still striving to grab a bigger bite of the cake.
PC vendors regard low-cost notebook as the first PC for consumers in emerging regions. But as prices of the new-generation low-cost notebook are no longer that "low", penetration in these regions is hard to expand.
Whereas for mature markets, most vendors regard it as consumers' secondary or third PC due to the relatively weak specifications (e.g. smaller screen size with lower-end components). Constrained by the physical limitations, low-cost notebooks that support 10- or smaller screen size will not be consumers' prioritized PC for work, from the perspective of both vendors and consumers.
Therefore, low-cost notebook is more like a segment for sub-PC users – those consumers who have a stronger demand for Internet connection. In light of this, many PC vendors have forged partnership with telecom service carriers.
For example, Asustek has bundled its Eee PC with a leading European telecom service carrier's 3G connection service, aiming to boost sales to top 2mn units in 1H08. Rival Acer is also in talks with Europe- and US-based telecom service carriers for dual-brand product launch, with products slated to be available in 2009. HP, at the meantime, has also discussed with Taiwan Fixed Network and Tatung 3C on the launch of a three-year NT$688 monthly package plan for a mini notebook and broadband connection package.
As Internet connection penetrates alongside with a shrinking notebook ASP, PC sales undergoes a fundamental change. More PC are being sold in bundled package (PC plus Internet connection), thus spurring consumption interest for those who are looking for Internet connection but without a solid PC consumption power, i.e. potential PC users. As most vendors set their package duration for three years, this also matches life cycle of notebook and encourages sales thereafter. Thus, having low-cost notebook to bundle with service from telecom service carriers somehow propels a new niche segment.
This niche segment has been successfully developed but a concrete market size is yet to be justified. Based on leading branded vendors' estimate, low-cost notebook shipments are expected to hit 12-13mn in 2008. But as the market is still under trial stage with upstream component suppliers being cautious over supply and PC vendors' cross-sector partnership is still in its infant stage, the shipments amount are expected to be 8mn. |
