Tiny Chip, Giant Ambition

Steve Sanghi has turned Microchip Technology from a near-dead maker of commodity processors into the top cloud seeder for next-generation electronics.

By Bob Parks, Business 2.0 Magazine

(Business 2.0 Magazine) -- Hands on" doesn't adequately describe Steve Sanghi's impulse for tinkering - whether it means donning a bunny suit at his company's chip-manufacturing plant to help troubleshoot defects, mixing it up with student inventors at science fairs, or outfitting his six-bedroom, 14,000-square-foot home on Camelback Mountain in Phoenix with custom upgrades. A few years ago, the lanky 51-year-old stayed up late in his garage building a series of indoor ramps for his Segway, an ever present companion at home and the office. When his basement water heater exploded recently, Sanghi asked the plumber how to sweat new copper pipes. With spools of solder and propane torches, the two finished the job together in a few hours.

Back at his drab office at Microchip Technology headquarters in Chandler, Ariz., Sanghi, the company's longtime CEO, is lately drawn to tinkering on a far grander scale: plotting with his customers about how to make his chips the star components of a vast array of new consumer and industrial products - wireless air-pressure monitoring for cars, office-energy automation technology, even silicon-powered running shoes. At first blush, that would seem an almost ludicrous notion, since for more than a decade Microchip has sold a 1970s-vintage processor, called a microcontroller, that's so crude and sluggish compared with the most mediocre laptop chip sold today that industry giants like Intel (Charts) largely abandoned the market years ago.

Yet Sanghi, more than anyone else, has turned the lowly microcontroller into a powerful new asset - not just for his company but for a new generation of inventors and entrepreneurs. While big-chip rivals like Intel and AMD (Charts) are still ruled by Moore's Law and ever more powerful processors, Sanghi tacked away from the semiconductor fleet more than a decade ago, betting that so-called embedded technology - linking smaller chips together instead of relying on high-speed workhorses like the Pentium - would help seed a wild variety of products with radically new capabilities outside the universe of PCs.

That dream is finally becoming a profitable reality. Serial inventor Dean Kamen relies on Microchip chips to run the electronics backbones not just in Segway scooters but also in his gyro-powered wheelchairs. The new all-electric roadster from Silicon Valley's Tesla Motors does zero to 60 mph in four seconds with help from more than a dozen Microchip processors that "charge-balance" power from the car's lithium-ion batteries. At Wynn Las Vegas, the $2.7 billion resort that opened in 2005, the centerpiece attraction, called Lake of Dreams, is a 20,000-square-foot lake illuminated by more than 4,000 LED fixtures, each controlled by one of Sanghi's chips. And Adidas's popular running shoe, the $250 Adidas-1, uses a single microcontroller to electronically stiffen or soften the sole. "It's an incredible family of products," Kamen says.

And an arguably incredible company. Once given up for dead as the No. 20 microcontroller maker in the world, Microchip has since eclipsed Intel and Motorola to become No. 1 in a $5 billion market, with sales growing at a healthy clip - 63 percent since 2002 - and profit margins consistently among the highest in the industry. "Microchip has an unusually diversified business for a semiconductor company," says Tore Svanberg, an analyst with Piper Jaffray. "With PC chips, most of the business depends on one big customer, Dell. But Microchip set out to sell chips for anything that's out there."

As avant-garde as Sanghi seems now to the rest of the chip industry, it wasn't long ago that he looked like another ladder-climbing lifer at Intel. After moving from India to earn a master's degree in electrical engineering at the University of Massachusetts, Sanghi joined the company in 1978 in its "special products division," which built specialized memory chips. For nearly a decade, he worked his way up, earning a reputation as a stern, often brash taskmaster - even in his dealings with higher-ups. When legendary co-founder Andy Grove interviewed Sanghi for the division president job in 1987, the newly appointed CEO casually asked Sanghi how old he was. Most young execs who had ascended the ranks so quickly would have been proud to respond. Not Sanghi, then 31, who bristled at comparisons with older colleagues. Above all, he wanted to project confidence. "What does age have to do with anything?" Sanghi shot back.

Grove didn't press further, and Sanghi apparently said all the right things. Two days later he became one of the youngest senior execs in Intel's history, running a $180 million-a-year division. "Steve was very sharp, very focused," says Boaz Eitan, a former Intel colleague who now runs Israeli-based Saifun Semiconductors. "A lot of people interpreted that as arrogance, but I didn't see it that way. High-output people can seem impatient."

Talk about impatient: Sanghi nearly doubled the division's revenue during the next two years and, just at the dawn of the PC boom, felt it was already time to start amending Moore's Law, the religion of Intel and most of Silicon Valley. The law predicts that chips will double in complexity every two years, but Sanghi believed that the value of pure computational power was overrated outside the realm of computers.

In Intel's automotive-chip division, for example, executives were pushing carmakers to place a single microprocessor near the engine block to control the vehicle's subsystems and accessories. Sanghi believed it made more sense to sprinkle simpler, cheaper, lower-power chips throughout the vehicle: one for the fuel pump, another for the carburetor, one each for the antenna, windows, and so on. That way, various parts makers could fashion products that were smarter on their own than they would be as slaves to a master chip. While PCs thrived on increasingly powerful processors, Sanghi thought there were hundreds of other products and devices that would benefit from being integrated with smaller and relatively dumb chips.

Sanghi, of course, wasn't wrong - just a decade or so ahead of the curve. (Today a typical car has about 30 microcontrollers.) He quit Intel in 1988, calling it "too bureaucratic, too political." Then, after a brief stint at a semiconductor startup, Sanghi accepted an offer in 1990 to become vice president for operations at Microchip Technology, a company that had just been spun out of 50-year-old electronics giant General Instrument.

Microchip sold the same type of memory chips (known as erasable-programmable, or EPROM) that Sanghi had developed at Intel, and he at last had center stage to play boss - but the good news ended there. Sales were down $37 million from $111 million a year earlier, and the company had just taken a $4 million loss. Even worse, quality control was a disaster (Sanghi discovered that 72 percent of chips were defective), execs began jumping ship, and memory-chip prices were stagnant, thanks in part to - what else - Moore's Law. Facing a financial meltdown, Microchip's board finally agreed to sell the company to the highest bidder, a Taiwanese chipmaker that offered just $15 million. But even that desperation tactic failed: Taiwan's stock markets crashed that fall, and the buyer killed the deal. When Microchip CEO Don Sorchych resigned three months later, Sanghi officially took the helm. "Steve didn't seem ready to be CEO," recalls Pierre Lamond, one of two partners at venture capital firm Sequoia Capital who had brokered the General Instrument buyout. "But who is?"

With the company's traditional EPROM business in ruins, Sanghi began looking for something - anything - around which to formulate a turnaround plan. He found it, eventually, in a then little-known microcontroller called a PIC. It was a leftover from the General Instrument era, sold primarily for use in keyboards and disc drives, that accounted for just 4 percent of Microchip's sales. Before Sorchych resigned, he had souped up the little PIC with a faster, simpler design and implemented a manufacturing technique called CMOS (now used in digital-camera sensors) that made the chips cheaper to make and less power-hungry when installed. Better still, they were less likely to become commoditized, since they worked only with a manufacturer's proprietary software.

While Lamond and others urged Sanghi to develop more complex (16- and 32-bit) processors, the new CEO was convinced that the 8-bit PIC could find a home in hundreds of products for years to come. "Steve arrived with a chip on his shoulder to show everyone he could do this," says Don Valentine, the other Sequoia Capital partner. "Pretty soon that chip became a 2-by-4."

Over the objections of Lamond, Valentine, and other execs, Sanghi turned the company inside out - jettisoning the EPROM business that accounted for the majority of sales and investing millions of R&D dollars into the 8-bit PICs that accounted for almost nothing. Sanghi had factory equipment retrofitted, reorganized departments, killed off extraneous product lines, reset employee benefits to zero, and eventually laid off about 600 out of 1,500 employees. "A lot of people bitched," recalls longtime employee Mike Finley, now an operations VP. "Few of us even knew what a PIC was when Steve showed up." Even Sanghi wasn't completely convinced of his plan. "If I had to go back, I wouldn't do it again," he says. "It was incredibly stressful. I aged 10 years in six months."

Yet the orders for PICs began to trickle in. By 1993, PIC sales had jumped eightfold, overall revenue had grown 21 percent, and Microchip posted a $4 million profit. But it wasn't until shortly before Microchip's $19.5 million IPO that year that Sanghi unveiled the weapon that would eventually catapult the company - which sat at No. 14 in the microcontroller market in 1992 - past No. 1 Motorola (Charts), No. 3 Intel, and other rivals.

Long before the open-source movement swept the tech industry, Sanghi decided to make PICs erasable and programmable, and he began giving away the software needed to customize them. While rivals sold read-only, or ROM, chips (for which customers would wait as long as a year to be programmed), Microchip marketed products far more in step with kitchen-table tinkerers. Engineers could buy their chips, have Microchip ship them out in just a few days, then load in their code. To Sanghi, the giveaway strategy was his only recourse against Intel's clout. "We'd be laughed at if we went into GE, General Motors, or Delphi," Sanghi says. "We had a nonstandard architecture coming from a small underfunded company. There was no way we could crack those markets."

But he could crack others with smaller customers - one reason Sanghi started spending more time throughout the 1990s at conferences with alpha geeks and less in sales pitches with suits and ties. "Strange as it sounds," says Tim Sanghera, a vice president at Mouser, an electronics parts distributor, "Microchip set out to create a cult around the 8-bit microcontroller."

One of the first converts was Dean Kamen, who had experimented with PICs to manage fluid levels in one of his early hit products, a home dialysis machine. During a meeting with Sanghi, Kamen pitched him on the idea of helping out at a series of robotics competitions for high schoolers. Sanghi jumped at the idea, and the two became fast friends. "Steve and I would walk booth to booth," Kamen recalls. "He knew the machines, knew the kids, he would be critiquing every part. He's a very hands-on techno geek - my kind of guy." More important, Kamen began giving the chips an even bigger role in the Segway, controlling speed and turning functions, managing battery power. And PICs will surely be a part of Kamen's newest alternative-energy applications, because the inventor loves the way the chips scale. "We stick with PICs," Kamen says. "Every bigger, higher-performance part ends up a natural extension of what came before. It makes development really easy."

Ihor Lys is another grassroots customer whose experimentation with PICs helped create what is now a $53 million company - and eventually helped move Microchip to the top of the industry. In 1997 the Carnegie Mellon Ph.D. formed a startup to sell LED lighting for use in upscale homes, theaters, and stores. Lys wanted to add advanced computer controls and the ability to dial up different color combinations, but the electronics would have made the fixtures too expensive. Recalling the Microchip products he'd used while working with robots at Carnegie Mellon, Lys created a prototype by writing a quick control program, burning it into a PIC microcontroller, and then soldering the chip to an LED. He and his partner showed the product to VCs a few months later, and soon afterward their startup, Color Kinetics, landed a $12.5 million investment. Today the company's PIC-powered lighting systems can be found in Brookstone stores and on the set of Oprah Winfrey's TV show. The Lake of Dreams built for Steve Wynn in Las Vegas is now the largest submerged LED lighting system in the world, with each of the 4,000 lights tethered to a Microchip PIC that monitors heat and relays instructions from a central server. Lys gives Microchip plenty of credit for his success. "PICs make our products so much easier to design," he says. "And they generate huge cost reductions, so we can spend more time thinking about how to make our fixtures better."

Even Sanghi's employees started catching the entrepreneurial bug. Ex-Microchip engineer Stephen Allen tinkered long enough with PICs to create a gadget that records all the keystrokes on a person's computer. He left Microchip in 2000 to start his own company, which now sells KeyKatchers and several other PIC-powered products.

Of all the budding markets for PICs, automotive still holds the biggest potential, since cars increasingly rely on smaller chunks of computing power spread throughout every component and system. Automakers rely on bigger rivals like Renesas and Motorola spinoff Freescale for advanced processors, but as with other applications, Sanghi is attacking the market through the side door: small inventors with big ideas.

Among them is Marc Tarpenning, the electrical engineering VP at Tesla Motors, a venture-backed Silicon Valley startup that in July unveiled one of the first all-electric sports cars, the $90,000 Tesla Roadster. Tarpenning uses Microchip PICs to handle battery management, power control, and constant data monitoring - all mission-critical functions. "You're running code for safety systems, and with microcontrollers we can go through it and double-check it in short order," Tarpenning says. "That would be impossible with a larger processor. You'd be talking about 1 megabyte of code vs. just 20K." Within three weeks of its launch, Tesla had orders for all 100 of its new cars. Perhaps it's no surprise that BMW recently inked a deal with Microchip to create new electronics for real-time tracking of energy usage and battery health.

Sanghi is also thinking far beyond cars. Microchip recently unveiled the first PIC product designed for the latest wireless networking innovation: the so-called ZigBee protocol that allows small chips to talk to one another with no wires, no central computing brain, and hardly any electricity. Companies like Eaton are already exploring ZigBee applications for medical data collection, smoke and intruder warning systems, and building-energy automation.

Relying on small buyers like Kamen and Lys might keep most of the Fortune 500 out of Sanghi's customer portfolio, but the Microchip model has some unique advantages. With 50,000 customers, no single buyer commands more than 2 percent of overall sales. Not being forced to dole out volume discounts is one reason the company commands a 60 percent gross profit margin. Staying small also makes it easy to keep indulging Sanghi's handyman instincts. When an employee got stuck recently with a dead battery, it was Sanghi who bolted out into the parking lot and jumped the car.


By focusing on the small but lucrative market for microcontrollers, Microchip has seen its stock and revenue soar.

Sales by product line

Note: Fiscal year ends in March. Sources: Bloomberg; CapitalIQ; Microchip Technology

Bob Parks is a freelance writer and the author of a new book, "Makers: All Kinds of People Making Amazing Things in Garages, Basements, and Backyards."


The company's processors play an important, behind-the-scenes role in a variety of innovative new products.


Ihor Lys, co-founder of Color Kinetics, credits the success of his $53 million company to Microchip, whose processors control Color Kinetics's giant LED display at the $2.7 billion Wynn Las Vegas.


Microcontrollers relay speed and turning-friction data in the Segway scooter; inventor Dean Kamen has also made the chips a staple of his gyro-controlled, stair-climbing wheelchairs.


Microchip processors help maintain precision battery control in the lightning-quick, all-electric Tesla Roadster.


The soles of the popular Adidas-1 running shoe soften or stiffen depending on the force with which you hit the pavement. A single 8-bit Microchip controller is the brains of the embedded electronics. Top of page

To send a letter to the editor about this story, click here.

Follow the news that matters to you. Create your own alert to be notified on topics you're interested in.

Or, visit Popular Alerts for suggestions.
Manage alerts | What is this?