How an Edinburgh Engineering Team Helped Build Sandisk Into a Global Powerhouse

It was the early 2000s and the dot-com crash brought an end to one of the UK's most hyped 1990s high-tech startups—Memory Corporation.

The Scottish startup rose to a dizzying market value with a clever approach: buying cheap, defective DRAM chips and using logic and packaging to bypass faulty memory cells.

DRAM was a precious commodity, and the repaired chips could be integrated into computer SIMMs (Single In-Line Memory Modules), extracting higher yields from existing chip manufacturing processes.

The market loved an irresistible prospect—the ability to create high value products from low-cost inputs.

However, the explosive demand for DRAM, spurred by the rise of PCs, led manufacturers to flood the market; Oversupply and pricing wars caused DRAM prices to plunge, and Memory Corporation's economic model collapsed.

As the DRAM opportunity fizzled out, the company reinvented itself under a new name—Torridon plc—inspired by the rugged terrain of the Scottish Highlands. Torridon formed a joint venture with the US electronics giant Solectron under the name MemQuest and pivoted its focus from DRAM to flash memory.

Much optimism was rallied around the new venture but then came the 2000 dot-com crash and dealt the final blow to a company boasting great innovation but facing unfavorable market conditions and unfortunate timing.

The Sandisk connection

Following the collapse, MemQuest's patent portfolio was acquired by flash maker Lexar Media. Sandisk, who was scaling into a global storage heavyweight, quietly seized the opportunity and recruited MemQuest's engineers.

Peter Smith, a Distinguished Engineer of ASIC Development Engineering at Sandisk, was one of them.

“Seven of us moved into Sandisk in 2002,” he said. “We had our initial suspicion that the company would make use of our technology knowledge and then we'd be dropped. But here I am nearly 25 years later,” he chuckled.

When Smith and the team joined, Sandisk was amid transitioning its flash memory from NOR to NAND technology.

NOR had proven to be a fantastic memory for firmware, but NAND was cheaper to manufacture in high capacities and at scale. And it would, indeed, become the memory format which drove the consumer electronics boom in the early 2000s—powering digital cameras, MP3 players, and the rise of smartphones.

“Our role was critical at the time,” Smith said, “because the algorithms we developed to manage NAND flash were significantly faster and higher performing than the algorithms Sandisk was bringing over from NOR.”

As NAND demand surged, Sandisk grew the team to capitalize on the opportunity. One of the first hires of the Edinburgh office was Kevin Kealy, Technologist of ASIC Development Engineering at Sandisk.

“Controller technology often creates the competitive advantage in flash products,” Kealy said. “The team that came from MemQuest developed this cool, innovative firmware and indexing approach that could execute flash management background operations far more efficiently. For consumers, it meant Sandisk CompactFlash (CF) cards could service a camera much faster than previously”.

By 2004, Sandisk's Type I CompactFlash cards were among the industry's most advanced, setting the benchmark for speed and capacity at that time.

Persistent State

Smith and Kealy have both been at Sandisk for over 20 years, and they know just how rare that kind of longevity is in the tech world. They credit an embedded spirit of innovation that has kept them both gratified and challenged throughout the years.

“You can't coerce a culture into a company,” Kealy said. “It's like trying to test quality into a product. You will always fail. Quality is something that must come from the ground up. And culture, too, must come from the roots.”

That spirit has been a driving force in their groundbreaking work on ASIC simulation.

The two can't share much about the platform's highly sensitive inner workings, other than its internal codename, in true Scottish fashion, is named after a single malt Scotch whisky.

What they could share, however, is that the virtual environment helps firmware engineers interact with an ASIC—the flash controller—that's in development.

The cost of taping out an ASIC is phenomenal. Nanometer sized features require highly complex masking techniques, pushing costs to millions, or tens of millions, of dollars and require months of manufacturing before a working prototype is ever seen.

“Clearly, you don't want to spend that kind of cash and build the wrong thing,” Smith said. “But it's also that you want to ship the product as soon as you possibly can to start recovering that investment.”

Simulations allow the firmware and ASIC to be developed in parallel, so engineers get a pretty good idea of how the prototype works before it physically exists. It's also crucial to getting products out faster, helping to better capitalize on market opportunities, and recouping the investment.

But hardware simulation is an inherently difficult field.

Even modest memory systems generate an astronomical number of timing and state combinations, while the complexity of flash management creates dependencies that are difficult to predict. Even basic physics can get hairy.

“The way we do it [simulation] is quite unique in the market and much in thanks to the work initially laid out by Peter Smith,” Kealy said. “The speed of execution that we can achieve comes down to some fundamental, powerful innovations.”

Up next: intelligence

This summer, Sandisk's Edinburgh R&D team begins a new chapter, moving from the docks of Leith to the heart of the city. The new office will sit within sight of the iconic Edinburgh Castle and near the city's world-class universities.

For Andrew Wilson, Senior Technologist of ASIC Development Engineering at Sandisk, it's the perfect mix of a beautiful setting with the hustle and bustle of the city center. But more importantly, it's an opportunity to bring the team closer to one of the world's richest ecosystems of scientific talent and bring young technologists—and fresh ideas—into the fold.

“Edinburgh is a world leader in various technical disciplines, from the Nobel prize-winning work on the Higgs boson quantum particle to the world's first cloned mammal, Dolly the sheep,” Wilson said.

“We've had fantastic success with recruiting local talent through our university intern program, particularly in applying natural language models to our simulation processes, and we're excited to keep building on that and continue to play our part in improving people's lives through technology.”

That, in essence, has been Sandisk's mission since day one. If there is one lesson from the company's history, it is that lasting success depends on more than breakthrough ideas alone—it requires technological leadership, strong business execution, and the ability to bring the right products to market at the right moment.

That is what made Sandisk a global powerhouse. Or, as they might say in Edinburgh: “aye, that's the business!”