The Past and Future of NAND with Deepanshu Dutta

Deepanshu Dutta always knew it was going to be chipsets for him, just not what kind.

“From day one, I understood that chips were the future. I didn’t really know the nuances of the semiconductor industry. I just knew that I wanted to work in Silicon Valley. I wanted to work on the future,” said Dutta, Vice President of Technology Development Engineering at Sandisk.

With a master’s in hand from UC Irvine, Dutta started his career at Sandisk in 2006, pushing the bounds of NAND technology with an irrepressible spirit for innovation. His work over the years has led to the development of more than 170 patents. Now, as a people leader and accomplished engineer, he’s poised to help guide Sandisk into a new era of excellence.

The halcyon days of NAND

The mid-2000s were golden years for 2D NAND. Between technological breakthroughs like TLC and a booming consumer electronics market, spurred by the introduction of the first iPhone, it was a time of rapid growth.

“When I got hired, we used to talk about a 50 to 60% bit consumption increase year-over-year, compounding every year in an exponential curve,” Dutta said. To meet the rising tide of demand, engineers like Dutta stepped up to the plate.

“I’d come to work every day and think about how to pack more information into the same physical space, these minuscule memory cells. That was the mission: smaller, denser, faster,” said Dutta. This is where Dutta’s innovative engine started.

“The average person who buys an SSD, they’re only focused on capacity, speed, and cost. MLC, TLC, QLC, these mean nothing to them, but they are the very reason why storage prices have come down significantly over the last 20 years,” he said.

Each generation of NAND technology presented unique challenges. The jump from MLC to TLC, and even now to QLC, requires novel solutions at a steady clip. It's not enough to fit another bit into the same memory cell and call it a day: speed, reliability, and ever-smaller form factors demand constant innovation. These are the challenges Dutta works on every day, the kind of work that energizes him.

“I love solving whiteboard problems. I love talking with my team about exciting, innovative ideas. That’s when I have the most fun at work,” he said.

New challenges, old ways

Those early years—as he racked up patent after patent—taught Dutta about the infinitely malleable nature of NAND. More bits packed into the same space, layers and layers of wafers, stacked in three-dimensional storage towers. These were lessons in the four vectors of NAND scaling. These innovations have already revolutionized consumer electronics, and now they’re the bedrock of AI data centers in the 2020s.

“GPUs are the heart of AI data centers, of course, but they need data to process. We’re talking petabytes and petabytes of data, and the amount of data is increasing by 20 times in some cases,” said Dutta.

As with all NAND development, the usual suspects are at play here—capacity, speed, and reliability—but there are new opportunities, too. Expanding into new and existing markets with QLC, closing the gap of storage and computation, or rethinking the underlying technology that powers LLMs and their massive data centers: these are the ideas Dutta and his team are focused on.

“We’re working on technologies that don’t just help our existing markets grow, but technologies that open entirely new lanes for the business. New hardware, new paradigms, it’s a very exciting time for us and NAND,” he said.

Dutta is confident, and he has every reason to be. The scale of the problem is immense, but the underlying challenges remain largely the same. Maximize capacity, minimize dead time for processing units, and enable reliability. These naturally feed into the myriad problems that AI data centers need to solve; higher capacity and more efficient drives will minimize the amount of energy needed to operate the necessary digital infrastructure.

While the ecological challenges add extra pressure to the technical work, Dutta and his colleagues have all the tools they need to take them on. After all, they’ve spent nearly 20 years perfecting those tools.

Meeting the promise of tomorrow

While Dutta concedes that the moment is unprecedented, his confidence is unmistakable. The innovative spirit that defined his career is his guiding light for the future.

“A technology company lives on innovation. If you don’t innovate, you die, literally. To survive is to innovate. We don’t think of innovation as just patents, just some side part of our jobs. It is the job. If we don’t innovate, who will?” he said.

This is a tall order, but one that Dutta feels his team and the company are prepared to face. From his perspective, the expertise and culture that permeate Sandisk, from Penang to Milpitas, will be more than enough to meet this moment.

“We’re constantly asking our customers what they need, we’re constantly benchmarking ourselves, and we’re constantly working together in our teams and with partners like Kioxia. We do this not to keep up with the competition, but to lead it,” said Dutta.

In the end, though, no amount of marketing research or client calls outpaces hard work and passion for solving the toughest problems. As a leader, Dutta is spending more time managing people and less time finding technological solutions, but his heart is still in innovation.

“I really believe that the most important quality for innovation is self-motivation. It takes curiosity to solve the hardest problems, and passion to do the hard research and development,” he said.

Dutta innovates for the love of invention and discovery, a passion that underpins Sandisk’s legacy and future.