And Intel said it will pull the Woodcrest server launch into this quarter.
Otellini was speaking at a financial analyst's spring meeting with every part of Intel involved in the restructuring. The analysis will include examination of weak business units and cost per unit.
Otellini also talked about Merom, Conroe and Woodcrest, and said that the firm is moving rapidly to a next generation design which will include a threefold performance per watt boost.
Merom in August will coincide with the natural notebook refresh.
He said Intel will transform Xeon DP product line to Woodcrest to the tune of 75 per cent by Q4.
Intel had realised it couldn't go any further with the existing architecture. It has large teams working on the next two microarchitectures, he said. Every two years Intel will bring out a new microarchitecture.
Intel is shipping 65 nanometre processors now. Presler, Dempsey and Yonah are shrinks of 90 nano technology. Intel wanted to ramp quickly on that technology.
The new chips this year will use 65 nanometre but he said Intel will ramp to 45 nanometre swiftly using the Nehalem technology. Yes, Intel re-uses codenames. All future processors will be optimised for performance per watt, and will leapfrog on existing designs.
The chipset designs will be offset by half a generation. Intel made this change several years ago. The object is for Intel to have sustained leadership in microprocessor technology.
Intel's Viiv will deliver over $1 billion of revenue in its first year, said Intel. It outshipped Centrino in its first quarter, he claimed. Vpro will also deliver $1 billion of revenue in its first 12 months at least, he said.
Future handhelds will be always on - with the ultimate handheld devices having CE device price points, said Otellini. By the end of this decade Intel handhelds will consume only half a watt. Over 1,000 people at Intel are working on this project today.
Intel is now focused on clearing inventory of existing CPUs, and has started the company wide efficiency project. It will ramp the new platforms during the second half of this year, and focus on winning back market share.
Server offering inadequate compared to AMD
Anand Chandrasekher, on after Otellini, claimed that Intel's market share loss stabilised in the first quarter of 2006, and that his firm is positioned to regain market share in the second half of this year. Intel had lost market share and there was an inventory buildup. Its twin engines of growth - emerging markets and notebooks are still intact, he said.
Intel will "burn off inventory" that has accumulated during the second quarter. The inventory accumulated as a combination of customers demanding more during stock constraints, a moderation of market growth, and "down channel" inventory growth. Customers were taking more product from Intel than the market wanted. Intel was constrained for the bulk of 2005 so its customers carried a "longer pipeline", said Chadrasekher. That introduced "cloudiness" into the market.
The new products Intel is introducing will prompt its customers to "burn off" the stock. He said Intel will clear out the excess inventory of several million units during Q2.
Intel lost market share in the channel, in the retail market, and in the servers. The chipset supply impacted it in retail. During 2005 Intel allocated its chipsets, supplying the server market from the top, prioritised its corporate desktop supplies and prioritised its mobile chipset supply Its roadmap in servers was inadequate because AMD had a better product, said Chandrasekher.
He said that Intel will win back market share on chipset supply in the future. It had recruited third party vendors (read ATI) to fulfil the shortfall. The move to 65 nanometres will ease its shortages, he said.
Fixing chipset supply will allow Intel to win back share - AMD did not have a better product. Intel decided not to bid on all the business because it knew it could not supply enough. But retail cost more share than it had anticipated. Intel mobo supply to the channel also showed a shortfall last year, but it can now meet channel demand, he said. The third party ATI chipset helped fix the channel demand, he said. Channel customers still have a desire to do business with Intel and when there is competitive product, market share will come back.
Chandrasekher used articles by Hexus.net and Anandtech to say that Conroe designs were fab. But you will remember some questions were asked at the last Intel Developer Forum about the odd way Intel allowed these two hardware sites to "review" the Conroe chip.
Servers was Intel's weakest area, admitted Chandrasekher. Bensley and Sossaman will help it win back share, while the second half ramp of Woodcrest will assist Intel to claw that back. The second half will see a "market share rebound" in the second half.
Yahoo is evaluating Woodcrest as well as a "lot of other customers" evaluating the server chip. Those include customers in the financial services community. Market share won't be clawed back as fast in the channel and retail side, he said. Intel already has over 100 design wins for Conroe mobos, he said. One hundred designs for Merom are already ready, he claimed. Intel has 200 plus Viiv designs slated for this year.
Bryant does bean counter thing
Chief bean counter Andy Bryant said he wouldn't try and charm the analysts this time round, which drew laughs from the analysts. He quoted Andy Grove as saying: "Profits are the lifeblood of enterprise. Don't let anyone tell you different." He said he was showing more information than he'd really like to.
Intel will reduce 2006 spending by $1 billion, make a $300 million reduction in 2006 capital expenditure, and look at the headcount as well as non performing groups of the business.
In 2006 two things including the ramp to 65 nanometres and overheads including labour were rising, but labout will go down a bit next year. He said unit costs of microprocessors will fall during 2007.
Dual core unit cost starts relatively expensive but comes down and by the end of next year the average price of dual core will be the same as unicore. The penalty for dual core is short term.
Microprocessor inventories will be down in Q2 and Q3 and are under control, said Bryant. Chipset unit costs in Q3 will be slightly more expensive and grow through 2007, because of the Broadwater platform, which is an expensive product. But quarter by quarter the average cost of the chipset will fall. This year there will be margin pressure because it's being produced in a 12-inch fab.
Capital intensity costs are more than Intel wanted in 2006 and are likely to stay the same during 2007. The reason is the chipset business is growing.
Forty five nanometre capital equipment by node is a considerable investment in 2006. But Intel has also invested in 90 nanometre tech because it's putting efforts into moving chipsets to 12-inch wafers. Nevertheless, Intel fabs will be full during 2007. µ
Sign up for INQbot – a weekly roundup of the best from the INQ