Sunday, September 30, 2007

Movers and shakers in semicon; new fabs in Asia

While speaking with the Fabless Semiconductor Association, USA, some time back, I quizzed them on the major movers and shakers (or slips) among the top 25, and what are the reasons for those.

Qualcomm has broken into the top 10 for the first time. It's the first time in the history of semiconductors that such a thing has happened, and is probably a sign of the times ahead.

One reason for this growth has been -- Increasing foundry orders. A Digitimes article reported that Qualcomm will increase orders by 15-20 percent in the September quarter to meet projections of strong 3G handset sales. According to the article, sources at Qualcomm suggest that wafer starts per month in the December quarter could surpass 30,000.

There have also been reports of strong June quarter for 2007. Qualcomm’s $2.32 billion in June quarter sales represents 19 percent year-on-year growth. Record chipset volumes of 65 million were at the high end of 62 million-65 million guidance. UMTS chipset shipments were noted to have increased by 127 percent year-over-year and 79 percent quarter-over-quarter, with the quarter-over-quarter growth rate roughly 4 times the market growth rate.

Finally, chipset ASPs increased by 2 percent year-over-year and quarter-over-quarter, and are likely to remain stable.

Situation in Asia
FSA quoted the Strategic Marketing Association's Quarterly Spot Report - July. While there have been announcements in India by SemIndia, HSMC, Moser Baer, etc., for fabs, there is every likelihood of another fab in the eastern Indian state of West Bengal. The technology ministry announced plans for a yet to be named fab at the India Design Center. Another possibility is a fab in Kochi, South India, by the NeST Group.

In China, Strategic Marketing Association expects that eight new fabs will start construction this year. Two fabs have started construction in the first quarter (Hua Hong NEC and ProMOS), and two more have started construction in the second quarter.

Grace Semiconductor, now headed by former Infineon boss, Dr. Ulrich Schumacher, which opened its first fab in 2003, began moving equipment into the shell of Fab 2, which was built at the same time as Fab 1. The company is installing used equipment and plans to begin production in Q1 0f 2008. The company also plans to begin building a 300mm fab, perhaps as early as next year, although financing is said to remain an issue for such a project.

Also in China, IC Spectrum began building a 200mm fab in Kunshan, about 45km east of Shanghai. Using 0.35micron technology from Toshiba, the new foundry expects to begin volume production by the first quarter of 2009.

TSMC began production in the second phase of its 300mm Fab 14 in Tainan in the south of Taiwan. This $2.4 billion fab will start production at 65nm and move to 45nm in 2008. There are also plans for a 300mm Fab 14, Phase 3 at the same location.

Looking at capex
In 2007, companies with capital spending budgets of $1 billion or more (the Billion Dollar Club) will account for 77 percent of all capital spending. Most of these companies (13 out of 20) are memory companies. Nine of these companies are from Asia Pacific (South Korea, China, Taiwan and Southeast Asia). Together, they plan to spend $23 billion this year, more than half of what the Billion Dollar Club has budgeted for capital spending.

Tuesday, September 18, 2007

Challenges of selling used equipment in global secondary market

The recently concluded SEMICON TAIWAN had a very interesting session on: The Challenges of Selling Used Equipment in Global Secondary Market, by Michael Mihin, Global Account Manager, Broadway Engineering Services Teams Inc. (BEST).

According to Mihin, there are about 600 IC fabs worldwide in 2007 -- 47 in Taiwan, 42 in China, 35 in Korea, 19 in Southeast Asia, and 28 in rest of world, 182 in Japan, 89 in Europe and 158 in the USA. As a result, the challenges posed by used equipment exists worldwide.

Mihin put down these challenges as -- rebuilding OEM equipment to capabilities equal to OEM specifications; supplier response at factory location instead of single country location; providing capacity at a lower cost with recycled equipment; and project revenue to support local resources in country with fab locations.

As for the benefits meeting these challenges, those are said to be extending life of tools by refurbishing, lowering cost of ownership, local contact for warranty and maintenance, parts support with local partners, out of country tech support within 48 hours, and partnering with local resources to meet fab requirements.

Applications include rebuild company supplies refurbished tool, fab supplies tool and rebuild company refurbishes tool onsite at fab, rebuild company changes chambers on in fab tool to reflect new process requirement, and rebuild company supplies refurbished chambers and fab installs on tool. Time is the issue with 2X the time in fab.

Equipment opportunities exist such as providing refurbished equipment that is capable of meeting new equipment operating specifications, availability of donor tools as market requires, new equipment versus refurbished equipment value analysis, and refurbished parts versus new parts to complete the project.

As for the pricing challenges, these include dealing with refurbished tool that is 30-50 percent less than new, travel costs and expenses without margin adder, software upgrades without margin adder, rebuild field service engineer hourly rate less than OEM, and market price of donor tools influenced by supply and demand.

The 200mm tool supply is increasing, and the refurbish process takes eight to 12 weeks for most tools. Apparently, regional support/alliances are being developed. BEST offers USA phone support Monday to Friday, and parts support in 48 hours from USA.

Concluding, Mihin said that the world was operationally flat. Challenges were being met by alliances and consolidation of service providers. Legacy tool support was moving to certified rebuild companies. Finally, local capabilities were needed to meet language and communication requirements of a fab.

The challenges of selling used equipment could be a challenge in India too, as and when fabs come up, make no mistake.

Broadband hasn't grown as expected in India

Yes, I believe so! The numbers, if one were to contend with those alone, DO NOT meet the expectations. Broadband was and is considered to be the new paradigm of India. However, are we anywhere near whatever growth we have been expecting? Let's see the stats for the various telecom segments.

According to the statistics made available by the Telecom Regulatory Authority of India (TRAI), the total number of telephone subscribers was 232.87 million at the end of July 2007, and the overall teledensity had increased to 20.52!

In the wireless segment, 8.06 million subscribers were added in July 2007 and the total wireless subscribers (GSM, CDMA and WLL (F)) base was 192.98 million. The wireline segment subscriber base stood at 39.89 million, with a decline of 0.20 million in July 2007.

And what about broadband? For broadband (≥256Kbps downloads), the total broadband connections in the country had reached only 2.47 million by the end of July 2007. In fact, during July 2007 there was an addition of 0.05 million connections!

Let's go back a few months! Venkat Kedalya of Convergent Communications had pointed out in an article to CIOL that India was nowhere on course to reach a target of 9 million broadband subscribers by this year! India has a target of achieving 20 million broadband subscribers by 2010, which now seems to be highly ambitious and well, unachievable!

Allocation of frequencies for BWA (broadband wireless access) is the immediate need of the moment. There is a need to look at WiMax and broadband over powerline (BPL) as far as technology is concerned. Some folks have entered the IPTV domain, so hopefully, we will get to see some content over broadband.

Even TRAI has urged the government to boost broadband growth. One of its suggestions has been to ask BSNL and MTNL to adopt a franchisee model so that local players may use their copper cables and offer high-speed Internet services. Decisions need to be taken for allocating spectrum for WiMax as well as making the National Internet Exchange of India more effective.

TRAI said: "Only 0.47 million broadband subscribers have been added in first six months of 2007, which is far below the growth trend required to achieve broadband policy targets. This necessitated an analysis of regulatory and policy frameworks, and to formulate new approach necessary for rapid roll-out of broadband in the country."

TRAI also accepts that while the growth of Internet subscribers was satisfactory, we are seriously lagging behind as far as broadband is concerned. It adds: "The government should ensure availability of more number of Ku-band transponders to roll out broadband services through DTH platform and utilize Universal Service Obligation (USO) fund to provide subsidy for providing broadband services through satellite in remote and hilly areas."

I'm not really sure how all of this will help. You do need at least a PC to access the Internet services. Am not sure how many folks are still willing to invest in home PCs and broadband, given that watching TV is a favorite pastime. Broadband over cable TV has not been a success either. What are we doing about this?

Monday, September 17, 2007

Indian government announces policy to woo investments in semicon fabs

Better late than never, as the saying goes. The Department of Information Technology, Ministry of Communication and IT, Government of India, needs to be congratulated for coming up with the Special Incentive Package Scheme (SIPS)to encourage investments for setting up semicon fabs, and other micro and nanotechnology manufacturing industries in India!

The "ecosystem units" have been clearly defined as units, other than a fab unit, for manufacture of semiconductors, displays including LCDs, OLEDs, PDPs, any other emerging displays; storage devices; solar cells; photovoltaics; other advanced micro and nanotechnology products; and assembly and test of all the above products.

Just a week or two back, I was in conversation with some companies from Israel who were looking to develop business in India. Now, they, and others, have clear guidelines to follow. One of the companies, Nova Measuring Instruments Ltd, should feel happy that the definition of "ecosystem" includes assembly and test of products.

Nova develops, produces, and markets advanced monitoring, measurement and process control systems for the semiconductor manufacturing industry. Another well-known player, Tessolve, has been present in India since 2005 and would surely feel glad with the notification. At least, the media and others will take more notice of the company.

In Hong Kong, an ex-colleague and I used to cover OLEDs. When I first read about this technology back in the early 2000, I used to wonder whether India could have such a capability. Seems, it is now in a position to have OLEDs! I hope Lite Array (OLED) HK is watching and reading all of this.

Plasma display panels is another interesting line. The guidelines should interest LG, Matsushita, Sichuan Changhong Electric Co. Ltd, IRICO Group Corp. Panasonic, Asahi, Mitsui Chemicals, Nippon Electric, Samsung etc. Some of these firms are already present in India in one form or the other. It's just a matter of their being keen on developing PDP in India.

LCDs could be another big investment area. Taiwan's AU Optronics (AUO), Chi Mei Optoelectronics (CMO), Sharp, Samsung, as well as other biggies like LG, NEC, etc., need to be wooed.

It really excites me to see all the possibilities in front of India. If this goes on well, India would be in for a great ride in electronics manufacturing, and in the semicon space.

In the same context, the Bangalore Nano 2007, which will be held in December, could not be better timed. There should be a whole lot of companies looking to be present at this show!

India's now on the threshold of major initiatives in the electronics manufacturing space. Some semicon fabs will also come up, and the number of fabless companies should likely increase. Maybe, TSMC and Tower could oblige with some foundries too. Should all of this happen at the right time, we are in for exciting times.

Saturday, September 15, 2007

Last mile problem -- in transportation!

The last mile problem in telecom is an issue close to the heart of all those part of the telecom industry. What it means is, providing or at least trying to provide some connectivity between two points where no other connectivity is possible. That's how wireless in local loop (WiLL) evolved.

There's another last mile problem -- in commuting from one place to the other -- at least if the distance is hardly anything. Not everyone's blessed with cars, bikes, etc. Some people still have to depend on other means of transport -- such as autorickshaws, buses, etc.

Well, my friend Bhaskar and I got into the "last mile problem in transportation or commuting", while trying to reach The Eros International Hotel in Nehru Place from Panchsheel Enclave in Delhi. Not a single autorickshaw we stopped and asked to be taken to our destination was willing to drive us from Panchsheel to Nehru Place -- which is barely 1-2km away.

We stopped and asked at least half a dozen autorickshaws, all of whom refused and merrily went their ways. Their reasons for not taking the route were: no mood, I'm headed this way; I'm driving on only straight; need to refill gas (maybe true), not interested, if you want to go to Trans Yamuna -- welcome, and so on and so forth!

Finally, the one who agreed, only drove us from the Chirag flyover to Nehru Place -- about a km away -- charging Rs. 30, though he later settled for Rs. 25! This was daylight (evening light) robbery at its best! What could we do? We were getting late for an event!

Am sure there are several like us who've experienced such a plight. Seems all the autorickshaw folks are the same everywhere. It all depends on their mood and distance, and of course, the haggling for more money.

There are no short-distance buses either. Reminds me of all those times when I either took trams in some other cities of the world; and the best of them all -- those 16-seater mini buses in Hong Kong -- which ply point to point. The drivers never grimaced even if there was only one person -- me -- seated in the mini bus -- especially the route from Wong Chuk Hang to Aberdeen!

Wish we had those mini buses here to travel short distances! Otherwise, we will continue to suffer from this last-mile problem in transportation!

Saturday, September 8, 2007

Toshiba-Sandisk Fab 4 takeaways for Indian fabs

The recent news regarding the opening of Toshiba-Sandisk Fab 4, the latest 300mm wafer fabrication facility for NAND flash memory at Toshiba’s Yokkaichi Operations, in Mie Prefecture, Japan, may have missed the eyeballs of many of us.

There are lot of takeaways for those looking to set up fabs in India. HSMC and SemIndia are building fabs in FabCity, Hyderabad.

Obviously, funding was never a problem for Toshiba and Sandisk. Toshiba funded the construction of the Fab 4 building, and Flash Alliance, Ltd, a Toshiba-SanDisk venture established in July 2006, 50.1 percent owned by Toshiba and 49.9 percent by SanDisk, is funding the advanced manufacturing equipment now being installed in the fab.

HSMC and SemIndia would have partners for its fabs. HSMC has roped in Infineon as a technology partner and licensed its 130nm technology, while SemIndia has AMD as a partner, besides having folks such as Flextronics and Broadcom, among others by its side.

Some other learnings from Toshiba-Sandisk's Fab 4 are: it has been designed to minimize any impact on operations from natural disasters. Also, Fab 4 would employ 56nm process technology at start-up, and plans for the gradual transition to 43nm technology, starting from March 2008. Maybe, the proposed Indian fabs would need to look at both aspects. Yes, 130nm is a good area to start, but let's look ahead at what the industry's doing too. By the way, Toshiba-Sandisk Fab 4 was built in about a year's time. So, the fabs coming up in India should perhaps, look at similar timelines.

May I mention here that recently, a member from an Israeli semicon delegation mentioned that it would not be a bad idea for India to have 200mm fabs, which would be able to make products for the aftermarket. However, India does not have a concept of aftermarket yet.

In between, there's news that Hynix signed a contract to sell the equipment in its 200-mm fab in Wuxi, China to China Resources Holdings. Now, should this have any bearing on the idea of a 200mm fab? Or should Indian fabs focus on having 300mm fabs? It's for them to decide!

It's worth adding here that the India Semiconductor Association (ISA) has been doing great work in making lots of things happen, the latest being the MoU with the Taiwan Semiconductor Industry Association (TSIA). The MoU focuses on developing business ties and exchanges between companies in the semiconductor industry in the two countries.

Slow India has already missed the Intel bus. Also, there are TSMC and the others, like Tower, who have an eye on India. Should they choose to open shop here big time, there's every chance that they might steal the thunder from these proposed Indian fabs.

If we are to be a world-class semicon hub, we simply cannot afford to be slow -- in strategy and in its implementation. Decisions should be quick and effective!

Tuesday, September 4, 2007

Indian semiconductor market growing thrice as fast as global market

As per the India Semiconductor Association (ISA) and Frost & Sullivan (ISA-F&S), India's 2007 annual growth in the semiconductor market is nearly triple the rate at which the global semiconductor market is currently expanding.

The actual total market (TM) was $2.69bn and total available market (TAM) was $1.26bn. By 2009, the TM will likely grow at a CAGR of 26.7 percent to $5.49bn and the TAM will grow at a CAGR of 36 percent to $3.18bn.

Anand Rangachary, managing director, South Asia & Middle East, Frost & Sullivan, said: "The global semiconductor total market is growing at a rate of 8-9 percent CAGR, whereas the India total market is growing at 26.7 percent CAGR till 2009. India, which represented 1.09 percent of the global semiconductor market in 2006 will be 1.62 percent by 2009. As domestic demand for all electronics products is growing India is emerging as one of the fastest growing region in the world."

India is one of the fastest growing regions in the world. TAM growth rate at CAGR 36 percent, compared to 26.7 percent of TM CAGR signifies higher growth in local manufacturing of electronics products. In the industry, the technology change is so dynamic that every year, a new application/product gets launched, which changes the demand forecast by many ways (eg. launch of iPOD or iPhone/ WiMAX/GPON/LCD TV) as well as ASP changes.

The government rules change demand, and therefore ISA captures these changes on a real-time basis. Hence, ISA decided to have an annual update of the India market report. All of these changes are well captured in the current report and India's growth looks almost three times compared to the global growth rate.

According to the report, the top five end-user products that are likely to drive growth are mobile handsets, desktops and notebooks, GSM base stations, set-top boxes and energy meters. Microprocessors, analog, memory and discretes are said to be the top four semiconductor products likely to drive revenues.

While these stats read great, I am wondering exactly how much of these handsets will be made in India. Rather, what percent of silicon going into these handsets will be made in India! Memory is said to be a driver of the revenue. Well, the DRAM market has been acting up.

Now, iSuppli reported in a recent report:

"Following a brief respite, market conditions for DRAM suppliers are set to take a turn for the worse in September, iSuppli Corp. predicts.

iSuppli previously forecasted that DRAM prices would undergo a downward correction in October, following the current period of relative strength that brought an end to a phase of severe erosion in the second quarter. However, iSuppli now believes the DRAM prices will begin to decline one month earlier, in September.

Near-term market conditions remain in a state of flux with a great deal of uncertainty in the supply chain as suppliers and distributors continue to work off a glut of DRAM inventory. Furthermore, sales momentum is waning in the DRAM spot market, as rising prices and falling supply of LCD panels cut into the available budget for memory in some PCs.

This is bad news for memory suppliers, which had been basking in the present period of relative pricing strength. Weak pricing in September will set the stage for further erosion in the fourth quarter. iSuppli now foresees the possibility of double-digit sequential price declines in the fourth quarter, erasing any increases that aided suppliers in the third quarter. Because of this, DRAM suppliers’ profitability will dwindle in the fourth quarter compared to the third, iSuppli predicts."

I'd be very interested to see how much of these memory predictions turn out to be correct! If they aren't, I wonder how memory is going to figure among the top four revenue drivers in the Indian semicon market, at least in the near and immediate future. Unless, I somehow missed a point somewhere!!