Why You Missed Your 2014 3D Stacked DRAM Equipment Shipment Forecast

Remember SEMICON West 2012?

How could you forget?

It was all clubbing with your customers at The Redwood Room, DNA Lounge, and The Endup after a few standout meals with them at Mission Chinese Food, Slanted Door, Wayfare Tavern, Acquerello, and Absinthe that were preceded by a few drinks with the team at House of Shields, Rickhouse, and The Old Ship Saloon that followed a day on the show floor that started after your solo morning run along the Embarcadero.

Yeah, that SEMICON West 2012.

And remember the CY2013 and CY 2014 shipment commitments you made for your 3D Stacked DRAM processing equipment at the annual sales meeting your company held in July 2012, just before the show?

Yeah, that equipment shipment forecast.

Your CEO needed some outside-the-box growth to keep The Street wagging docilely on its leash. You had some new, compelling 3D IC / TSV processing equipment freshly rolled out, most notably the 3DrICer™,* your latest and greatest, and you as product manager knew you were going to be enjoying some healthy organic growth shipping production 3DrICer™ tools into the MEMS and CMOS Image Sensor markets, where the tools were already qualified and in use.

But organic growth alone doesn’t keep The Street’s tail in happy motion and, along with an attention-grabbing acquisition surprise or two your CEO was planning to announce at SEMICON, it was time to belly up to the bar during the sales meeting and commit to equipment shipments for 3D Stacked DRAM fabrication.

Why? Well, as your CEO had read in some of the executive summaries floating around from market research firms covering the 3D IC / TSV space, word in mid-2012 was that 3D stacked DRAM and 3D Logic SOC applications would be the biggest drivers for the volume adoption of 3DIC technology over the next five years, followed by CMOS image sensors, power devices and MEMS.

The wave your CEO wanted to ride to the bank (her bonus being based on your stock price / market cap, among other metrics) was going to be powered by those big drivers. She (your CEO) had the market research data to trot out for the analysts as part of her story, but that wasn’t quite enough. That’s why your CEO needed an aggressive equipment shipment forecast from you, in mid-2012, for 2013 and 2014 equipment shipments into the 3D Stacked DRAM market and, despite your gut, that’s what you delivered, and that’s what your CEO promoted in interviews at the show.

Based on the analysts’ forecasts for global shipments of something like 350,000 3D Stacked DRAM 12” equivalent wafers in 2014, growing to 600,000 12” equivalent 3D Stacked DRAM wafers forecast to be fabricated in 2015 (from a base of ~0 in 2012), you worked with the sales team to produce your 3DrICer™ Stacked DRAM equipment booking and shipment forecasts for 2013 and 2014.

Working backwards, if the world needed processing capacity in place for almost 600,000 12” equivalent wafers by 2014 year end to have the installed capacity to deliver to the 2015 3D Stacked DRAM wafer forecast, then the world was going to need … at least one tool per DRAM manufacturer (if not two each; inefficiencies), in place and commissioned by early to mid-2014 at Samsung, Hynix, Elpida, Micron, Nanya, Winbond, Powerchip, and ProMOS, 2012’s usual DRAM suspects.

With a six month lead time to build and ship a tool, and with a twelve to eighteen month sales cycle, you pretty much needed to be talking orders on the SEMICON West 2012 floor for 3D Stacked DRAM tools to have the tools in place for the forecasted 2014 3D Stacked DRAM device shipments.

But it didn’t happen. You and your major competitor missed your 2013 and 2014 3D Stacked DRAM equipment delivery forecast. Why the disconnect?

It might have been because of this: There was an overall, and significant, DRAM market contraction starting from 2011 and extending through 2012. With the bottom falling out of the DRAM market, capital equipment investments by the DRAM players who were your target customers went the same way: South.

Particularly capital equipment investments for new technology and for specialized production equipment that would be used for DRAM products whose production was still on the horizon, or over the limb. From your customers’ perspective that would mean forgoing investments in products like 3D Stacked DRAM which, unfortunately for you and for your forecast, meant taking your 3DrICer™ forecast down with that ship.

A sinking bound to last at least until the rising DRAM tide lifted all ships again … probably not until 2014 or later, the DRAM market forecasters said in 2012.

Then there was this headline from earlier (27 Feb 2012) in the year: “Elpida seeks bankruptcy protection, $5.6 billion debt.” “Elpida Memory Inc filed for protection from creditors on Monday with $5.6 billion in debt, the biggest bankruptcy filing by a Japanese manufacturer, after potential partners failed to come through to rescue the cash-strapped chipmaker.”

Elpida was arguably the industry’s biggest cheerleader for 3D Stacked DRAM; for example, it was reported in 2009 that “Elpida Memory recently pushed vertical stacking of DRAM to new heights by connecting eight 1G chips using through-silicon vias (TSV), creating what it calls the world’s largest-capacity DRAM with ~8GB of storage. … A manufacturing line has been set up in Elpida’s Hiroshima plant, using Akita Elpida’s package processing technology, the company said; the Nikkei Business Daily pegged this as a ¥5B investment for a 10,000 wafers/month TSV line.”

Without Elpida’s enthusiastic cheerleading, where was commercial fabrication of 3D Stacked DRAM to receive its push?

How was 3D Stacked DRAM going to get its game on? From whence was 3D Stacked DRAM to regain its mojo after Elpida’s 2012 collapse, and with that mojo regained you meeting your equipment shipment forecast?

And, finally, there was this: In 2012, the ink was still drying, if there was ink at all, on the important industry technical specifications for 3D Stacked DRAM in both its flavors – Wide I/O, and Hybrid Memory Cube.

Without JEDEC standards (for Wide I/O DRAM), or without codified standards from the HMC Consortium (for Hybrid Memory Cube), being in place for 3D Stacked DRAM products in 2012, DRAM-using customers in the data center or smartphone spaces, for example, would be at great risk if they were to prematurely design 3D Stacked DRAM components into their next-generation products.

In 2012, the 3D Stacked DRAM cart was definitely out in front of the industry horse.

As was your 3DrICer forecast for 2013 and 2014 tool deliveries.

Ouch.

But you hung tough and stayed loose, and in Part 2 of this piece, “What’s Different for 3D Stacked DRAM Equipment Shipments in 2015,” have reason to enjoy a renewed outlook for shipping your equipment.