Despite LSI Logic's announcement earlier this week that it would be leaving the budding structured Asic market, analysts believe the sector still has a future.

Representing a further sharpening of the company’s overall strategy, LSI Logic announced on Monday it would be focusing the company on storage and consumer related markets represeting a shift away from being a broad-based Asic supplier.

As a result, LSI is discontinuing RapidChip development, according to CEO and president Abhi Talwalkar during a conference call.

Bryan Lewis, v-p and chief analyst for semiconductor research at Gartner Dataquest told Electronic News of LSI’s move that, “Clearly it is a major impact on the structured Asic market.”

At the same time, he made clear that this move doesn’t mean the structured Asic approach is dead, and is revising his forecast to reflect this change.

“This was always a niche but it is clear it was nice in terms of design starts with lower NRE. But big customers control the market are not as concerned about NRE – they want performance with smaller die size,” he continued.

Previously, Lewis predicted structured Asics would grow from one per cent of total Asic revenue in 2005 to five per cent of total Asic revenue in 2008, noting that current market data suggests 2005 will come in slightly below the forecast.

Specifically, Gartner has reported the structured ASIC segment would grow from $100m in 2004, to $239m in 2005, reaching $1.5bn in 2008, he said.

This is against the back drop of the $21bn total Asic market last year, expected to reach $27bn in 2008.

Structured Asics go after the FPGA, but the big money is in cell-based, he believes.

Lewis also admits this may have a negative impact on the market perception, but if anything will give Altera some business, which he predicts was approximately $50m in 2005 for its structured Asic HardCopy service.

Altera reported last year approximately four per cent of revenue came from HardCopy, which it will aim to increase over the next few years to be between 10 and 15 per cent of total revenue.

Gartner views the structured Asic methodology to fall under a larger, platform Asic umbrella that includes all array-based methodologies that are customised at the final mask layers. Platform Asics also include cell-based platforms that don’t save on mask charges as structured Asics do, but design reuse is high, Lewis explained. Cell-based platforms include Texas Instruments’ OMAP, Philips’ Nexperia, as well as others by companies like Panasonic.

“The platform as a methodology is not dying by any means. The structured Asic as a value proposition is delivering, but this has put a rifle shot in the structured Asic camp,” Lewis noted.

“There is a place in the market for structured Asics, but not the mainstream,” he concluded.

LSI’s main competitor in the structured Asic market Altera’s Paul Hollingworth, senior director of the company’s HardCopy product group sees, “a great deal of change for structured Asics because this is an emerging embryonic market.”

Altera definitely perceived LSI as its biggest competitor in this space, so Hollingworth said, “On one level, we are obviously delighted and any of their customer that are feeling let down, they should feel free to call us.”

Now however, he noted, one of the obvious questions is, “Does this mean the market does not exist?”

But even Talwalkar made it a point during an early morning conference call to say that structured Asics do deliver on their value proposition.

Still, Hollingworth said Altera wishes LSI well in their new direction, but at the same time says this is the end of an era of LSI as a broad-based Asic supplier. “I was very surprised,” he admitted, but “companies need to move with the times.”

“If you look at some of the causes, when LSI went into RapidChip, it was a bold and aggressive move. They didn’t invest in a back end flow specific to structured Asic, but used their standard cell flow, which they even said was effort-intensive. That approach doesn’t scale and you have to cherry pick designs,” he explained.

Altera approaches this from a fundamentally different angle, and its tool flow is meant to support very large numbers of customers, he said.

“Our back-end flow is engineered around minimal support being required, with maximum ease of migration. I think we’ve come at the problem from a very different point with a lower average cost per design,” he added.

To Jordan Selburn, principal analyst at market research firm iSuppli this is pretty big news.

“We have felt and were confirmed by [LSI’s announcement] that the revenue ramp to structured Asics is not much faster than for cell-based Asics even though the development of the chip may be much faster,” he said.

“If in fact our revenue forecast is correct, what this means is that the market is not going to be big enough, fast enough to make a difference to the big Asic customers,” Selburn continued.

However, the market is big enough for one or two companies to get to 100 million this year, but if you are an LSI looking to turn your Asic business around, it may not be growing fast enough for you, he offered.

“Even a few years ago, [LSI] was talking about a $100m run rate for RapidChip. They felt there was a compelling market – and structured Asics fit the bill for a big chunk of the market that could not afford to go to cell-based Asics,” he explained.

The big question that comes up is whether this is an indictment of structured Asics as a whole.

Not at all, Selburn believes, but the situation may indicate that cell-based approaches at larger geometries still have a lot of steam left, he concluded.

www.altera.com
www.lsilogic.com
www.gartner.com
www.isuppli.com