You are here

Cypress Startups: History, Theory of Funding, Lessons | Cypress Semiconductor

Cypress Startups: History, Theory of Funding, Lessons

Last Updated: 
Aug 25, 2014

My 1992 book No Excuses Management, stated that Cypress planned to grow by funding internal startups. The idea was that when we reached $1 billion in revenue, Cypress would have avoided becoming a big, bureaucratic company, evolving instead into a “federation of entrepreneurs” that still had the just-do-it startup attitude. To date, Cypress has funded nine internal startups and acquired two others, Silicon Light Machines and SunPower. There was one mega-winner, SunPower; one big winner, Cypress Microsystems; two winners, Fab 2 and Fab 4; five flops and the jury is still out on Cypress Systems and AgigA Tech. Our process calls for Cypress to fund the venture as its sole venture capital source. Startup employees own stock in their own company, which has equity separate from that of Cypress. Success occurs when the startup is “spun in” (Cypress acquires the company by buying the employees’ stock) or “IPOed” (shares of the startup are sold in an Initial Public Offering (IPO), taking the startup public and providing liquidity to its employees). After an IPO, Cypress’s shares in the newly public company may be held to control the company, sold to raise cash or “spun out” (distributed pro-rata to Cypress shareholders).

Cypress bought out the employee shares in each of our startups, except for Silicon Magnetic Systems, which went out of business, SunPower, which had an IPO, and Ross Technologies and Silicon Light Machines, which were sold to other companies.

CYPRESS TEXAS INC. (CTI)1 FAB 2 12/1986 $41.9M Q2 1990 $4.9M 6.6% $79.6M $7.4M $1.00
ASPEN BiCMOS 8/1987 $15.2M Q2 1993 $5.2M -11.5% $16.2M $2.9M $0.20
MULTICHIP TECHNOLOGY (MTI) MODULES 3/1988 $4.0M Q2 1993 $5.5M 10.9% $17.2M $2.8M $0.60
ROSS TECHNOLOGY MICROPROC 12/1989 $18.0M Q2 1993 $4.0M -69.1% $21.8M $2.6M $0.34
CYPRESS MICROSYSTEMS (CMS) PSoC 12/1999 $55.7M Q4 2005 $12.0M 18.0% $232M $15.2M $0.65
SILICON LIGHT MACHINES (SLM) OPTICAL 8/2000 $115.2M Q3 2008 $2.6M 15.4% $11.0M $0.0M $0.37
SUNPOWER (SPWR) SOLAR 5/2002 $142.8M Q4 2005 $29.3M 5.1% $1.1B $141.7M $18.00
SILICON MAGNETIC SYSTEMS MRAM 10/2001 $36.9M Q1 2005 $0.0M N/A N/A N/A N/A
CYPRESS SYSTEMS MODULES 2/2007 $8.6M Q3 2010? $10.0M? 22%? $100.0M? $15.7M? $1.00?
AGIGA TECH NV MEMORY 3/2009 $5.8M Q3 2011? $12.5M? 30%? $99.4M? $2.6M? $1.00?

  1. The target pretax profit for our fabs was 0%—to sell wafers at cost.
  2. The “buyout” for CMI was the continuous vesting of Cypress options for CMI employees as milestones were met.


Our first startup was fab 2, formally Cypress Texas Inc. or CTI, founded in 1986 in Round Rock, Texas. We decided to make our second fab an independent startup to keep the lean mentality developed in our startup fab 1 in San Jose, which was very successful as an entrepreneurial organization. Fab 1 remained viable for 22 years, from its first wafer shipments in 1984, until its sale in 2006, from which we recovered the book value of all of the equipment and were able to preserve its jobs by turning it into a new startup, the Silicon Valley Technology Corporation. SVTC was chartered to do wafer process development for outside companies, a practice Fab 1 started to tune up its P&L statement. (All Cypress organizations are run as Profit and Loss centers—even our legal department.)

Fab 2 also remained viable for a long period of time, from 1986, when we bought a used magnetics building to house it, through 2008, when the fab was shut down due to obsolescence. We will recover about $20 million from Fab 2 by selling its equipment, building and land. We almost saved fab 2 as another spinout foundry, but a recession and a severe glut of wafer fab capacity hit the market at exactly the wrong time. The founding CEO of fab 2 led the organization to a successful spin-in acquisition in which Cypress purchased the employees’ shares in 1988. The follow-on VPs who ran fab 2 had good and bad years. One big Fab 2 problem was quality, an area where the lean startup mentality can be a problem. The second-to-last manager of fab 2, Minh Pham, brought and held the facility to a sustainable high-yield, high-quality, low-cost operating status.

The lessons learned from the Fab 2 experience: 1) Internal startups work. 2) Startups always cut corners on quality. 3) Internal startups—including all Cypress Business Units (BUs)—should stay continuously ready for “life after Cypress.”

Aspen Semiconductor was an outside group referred to Cypress by one of our venture capitalists, John Doerr of Kleiner, Perkins, Caufield and Byers. We funded Aspen to develop BiCMOS technology. The legacies of Aspen are the SM13 and B53 BiCMOS technologies that have produced some of our most profitable products, such as RoboClock and HotLink, which were designed in other BUs using Aspen technology. Aspen was successful to the point of producing $5 million per quarter in revenue (half of its target) but never broke even. We spun Aspen into Cypress for $0.20 per share (vs. the $1.00 per share target) shortly after its founding CEO was fired for violating Cypress’s honesty policy and for having taken proprietary information from National Semiconductor, his prior employer, causing Cypress to be sued. The only key employee of Aspen that stayed at Cypress long term was Tony Alvarez, who later became the VP of Cypress’s process R&D group and then EVP of our Memory Products Division. One big benefit of internal startups is that the economics of “founders stock” can be used to recruit some very good employees. Internal startups are also very useful in retaining employees being recruited by outside startups. “Why go to that flaky startup,” we can ask, “when you can join a startup within Cypress and keep your Cypress options, too?”

The lessons learned from Aspen: 1) Do very careful reference checks on the integrity and capabilities of the founding team. 2) Demand that the startups use your business processes unless there is a compelling reason not to do so. 3) Don’t invest in niche process technologies, such as BiCMOS, that cost almost as much to develop as platform technologies, but have very limited markets.

Multichip Technology Inc. was an outside group funded by Cypress in 1988 to address the Silicon Valley “hot technology” of the time, multichip modules. It was thought that companies that could solder together their chips into entire systems on small modules (circuit boards) would displace systems companies, which had uncompetitive manufacturing costs. We soon realized the folly inherent in the theory of trying to compete with systems companies with a few people in the startup company. That is obvious today, but it was not in 1988. Multichip was quickly degraded into becoming a memory module company. I considered that business to be a loser, calling the products “memory on a stick.” Nonetheless, Multichip did achieve revenue of $20 million per year with 10.9% pretax profit. We realized Multichip had created a module manufacturing capability that we wanted, but that it would never be a viable stand-alone company. We bought out the employee shareholders of Multichip for $0.60 and merged it into Cypress.

Multichip’s founding team and I parted ways when we continued to argue about the company’s charter. We founded Multichip to put Cypress chips into modules and thus sell our products at higher prices (the same premise behind Cypress Envirosystems). The president of Multichip made most decisions on a Multichip-first basis, sometimes buying memories from our arch-competitor, Samsung, including off-spec SRAMs that produced inferior-quality product. The management of Multichip misled me for several quarters into believing that it would begin to use Cypress products. I finally refused to sit on the Multichip Board any longer, leaving Cypress director Fred Bialek as the Cypress representative. Multichip disappeared during the next downturn.

The lessons learned from Multichip: 1) Having separate equity incentivizes the startup management team to make decisions that make their stock more valuable, even if it hurts the Cypress patent. 2) Do a careful reference check on the integrity of the founders. 3) Fire the CEO if he even once takes an action harmful to Cypress and “overlooks” telling you about it in advance. 4) Do more careful market research on the business plan.

Ross Technology was formed from an outside team lead by Roger Ross that I actively recruited out of Motorola to pursue the SPARC microprocessor business with our partner, Sun Microsystems. Along with MIPS, SPARC was an example of another hot technology, the Reduced Instruction Set Computer (RISC), that was “sure” to dislodge Intel’s older, less elegant microprocessors. Cypress and Sun produced the first integrated version of SPARC with a joint design team. I saw that our team was not capable of designing the next generation SPARC microprocessor, which featured a multi-threaded CPU and multiple cache memories. Among those I recruited to help Cypress was Roger Ross, the manager of an advanced microprocessor group at Motorola. I heard through the venture capital network that Ross’s reputation at Motorola was to impress technically but to fail ultimately in delivering fully working products. I should have listened to that advice. Ross did bring a talented Austin-based team to Cypress, for which we were sued for theft of trade secrets by Motorola. That suit was dropped when I made it clear we would fight back and win. Ross won a significant fraction of the second-generation SPARC business at Sun. When Ross’s revenue grew to $15 million per quarter with over 20% pretax profit, we assembled an IPO team to take Ross public in 1999.

Unfortunately, Ross’s culture of corporate dishonesty and poor quality did them in before the big event. Despite lecture after lecture from me on the criticality of designing and testing with full margins, Ross’s designers quietly cut corners—with the approval of the boss. I clearly remember the meeting at Sun Microsystems in which I personally carried over the latest group of impaired SPARC modules that needed engineering waivers, to ask if Sun would “try them out in the system.” They said, “No, thanks.” They weren’t interested in any more modules that didn’t fully meet their specs. Ross degraded rapidly after losing its biggest customer. We spun them into Cypress for $0.34 per share and worked diligently to get new customers. We eventually sold Ross to Fujitsu for $27 million, recovering most of our investment.

The lessons learned from Ross: 1) Do more diligent checking on the CEO and team—and then believe what you find, if it is negative. 2) Never permit the founder to name your startup after himself. 3) Immediately change out the management team if it is not completely honest (for every dishonesty you discover there are 10 you do not find out about until it’s too late). 4) You can’t compete against Intel with a startup. 5) When your company is the fourth or fifth pursuing a new technology, it’s a me-too startup that will not make it. 6) Even after a disaster, work hard to increase the salvage value of a startup.

Fab 4 was also launched as the startup Cypress Minnesota Inc. or CMI in 1990 with Lothar Meier, a long-time Cypress veteran, as the CEO. It was a great deal—we bought a fully functioning fab in Bloomington, Minnesota from the computer company CDC for only $17 million. Just as there was folly in Ross’s competing versus Intel, so there was folly in CDC’s belief they could simply jump into the brutal chip business by spending $100 million on a fab.

We solved two problems after learning from our prior startups. First, Lothar Maier, a Cypress veteran, was appointed president of CMI. We knew he would not make Multichip-like “them vs. us” decisions. Second, we incentivized the team at CMI with Cypress options (in excess of what Cypress employees were offered) to make sure CMI’s economic interests were aligned with Cypress. CMI was also run very lean, as a startup should be. It was very successful for us from its founding in 1990 until about 2005. Equipment companies considered Cypress to be the leader in fab asset management. The fab was complimented with a superb process R&D group that produced the lowest-mask count wafer processes with the lowest mask count in the industry (typically 20, versus our competitors’ 30). Thus from 1990-2005, we enjoyed the lowest wafer cost in the industry, although our costs were higher on a per mask basis than the best offshore foundries. CMI simply became part of Cypress gradually as the performance-based Cypress options vested. Around 2007, we began to realize that CMI, which by that time had been an internal corporate business unit for a decade, was no longer lean. It was living in the legacy of low-mask processes, while its per-mask cost—even ignoring depreciation—was not competitive with other foundries from which Cypress could buy wafers. This problem has wiped out the U.S. fabs of many companies, including LSI Logic and Freescale.

Today, CMI is fighting for its life in a crunch cost-reduction program. Its problems are exacerbated by the fact that there is an excess of capacity on the market driving down wafer prices and an economic recession causing underloading. Minh Pham has transferred from CTI to CMI to see if he can work the same turnaround there as he did in Fab 2. The initial finding is that the fab had gotten very fat, giving him hope that it can be made competitive again reasonably quickly.

The lessons learned from fab 4: 1) Given the lack of a continuing significant financial incentive to be lean, corporate organizations always get fat. 2) From CDC’s foray into semiconductors—it is very risky to stray outside your areas of core competence.

Cypress MicroSystems Inc. (CMS) was founded in 1999 to diversify Cypress by entering the microcontroller business. The Seattle team that created our very successful USB business (itself a startup, formerly the ICD clock company, acquired by Cypress in 1992) brought forth the unsolicited CMS business plan. The idea was to use our microcontroller-based USB technology to enter the microcontroller market—specifically to attack Microchip Technology, a leader in the microprocessor business that nonetheless had inferior silicon and design technology, relative to that used on our USB chips. To get many products to market quickly (our competitors had thousands of different microcontroller products), the team invented the Programmable System on Chip, or PSoC. It is now the flagship product line at Cypress. PSoC products were designed to be highly configurable for the “wrong” reason—to “steal sockets” from industry leader Microchip Technology by providing the same functionality as Microchip’s portfolio of 5,000 parts. The programmability of PSoC allowed us to design the equivalent of 5,000 Microchip products into only six PSoC programmable chips.

The socket-stealing idea did not work out because it meant we always had to lower prices, our only weapon to steal sockets. However, the extreme flexibility built into PSoC became favored by customers with new designs who could now change their hardware at the last minute, something a typical microcontroller did not allow. In other words, PSoC beat the competition in winning new designs, not stealing old ones.

The PSoC BU is currently producing about $150 million per year in revenue and growing. The CMS team in Seattle was rewarded by having their CMS shares bought out by Cypress for $15.2 million in 2003. In this case, the reward was less than the nominal share price in the business plan ($0.65 per share at buyout vs. $1.00 in their plan), because of share dilution—the startup used $54 million in cash before turning profitable, compared to the $20 million in their original business plan. CMS has obviously been a company-changing success for us. Most of the key employees of CMS are still with Cypress, pursuing their PSoC dream. The original CEO of CMS was removed, based on my learning from prior startups. George Saul took over to lead the team to victory.

We suffered through a four-year battle putting CMS products into Cypress manufacturing due to its poor quality. At the peak, we suffered 50 customer returns per quarter for “test holes,” which caused insufficient testing of the product. A corporate task force was needed to fix the problem.

The lessons learned from CMS: 1) If a startup gets stalled, change management—even if you like the CEO. 2) Our startup methodology—in which greater than expected losses do not completely wipe out employee gains—is superior to any corporate bonus plan for motivation. 3) Startups always cut corners on quality to make their numbers. 4) Startups work harder than big companies, even hard-working ones like Cypress.

Silicon Light Machines (SLM) was an external startup we acquired in September 2000 for $115 million (2 million shares of Cypress at a time when Cypress’s share price was over $50). It was the one time that Cypress’s Board allowed me to make a quick acquisition without doing a complete job on due diligence. I portrayed the acquisition as needing to happen quickly, based on a funding event at SLM. The fact is we could have done better due diligence and should have. SLM made optical MEMS (Microelectro-Mechanical Systems) that controlled light with electrical signals. By shining red, green and blue lasers on three SLM chips and combining images, the company created a projection television set with astonishing image quality—better than that of Texas Instruments, which was moving into the projection TV business at that time. We were more interested in the communications uses of SLM’s chips, thus we worked with SLM before acquisition to sell their projection TV technology to Sony for $75 million, which was paid at the rate of $4 million per quarter. SLM thus became an acquisition with optical capability and $75 million in hardwired funding (making the actual acquisition cost “only” $40 million). SLM was supposed to add the optical capability to Cypress that would give us end-to-end data transmission capability during the dotcom boom. In retrospect, that was a very shallow rationalization that never did work out. After I got more deeply into the optics required for data communications, it became clear to me that the SLM technology did not bring any advantage to the dotcom world.

We continued to sell the optical MEMS already developed by SLM, but in non-TV markets, as required by our Sony contract. That business grew to $4 million per quarter but never justified the acquisition of the company. SLM’s final project did use their optical capabilities to create a product for a market in which Cypress was engaged. The optical navigation system (ONS) is a technology to make an optical mouse with extraordinary resolution and capabilities. Since Cypress has a business unit that sells USB mouse chips, the ONS mouse chip was a natural companion. We successfully completed the ONS mouse project and achieved a few hundred thousand dollars in quarterly revenue. The ONS chip is now gaining traction in the “finger navigation” market (turn the mouse upside-down and run your finger over the sensor to create movement)—but the jury on the last SLM product is still out, nine years after acquisition.

The lessons learned from SLM: 1) Never make an acquisition without having a complete business plan that demonstrates exemplary ROI. 2) Once you realize that you have made a mistake with a startup, sell it or disband it immediately. 3) Straying outside of your core competencies is very risky.

SunPower Corporation was an external startup company that had stalled at $2 million per quarter in revenue. It was threatened with bankruptcy at the beginning of the dotcom crash. Cypress was one of the early adopters of solar energy in Silicon Valley. We did it in 2000, when solar cells were only 14% efficient, meaning that a roof full of them could produce only 35% of the power used by our new headquarters building. That put me on the lookout for a better solar cell. In a now-famous meeting that never happened (there was an intermediary), I became re-acquainted with my old Stanford school-mate, Dick Swanson, a genius who had dedicated his career to solar energy—and had created a 21% efficient solar cell that produced 1.5 times more power per silicon wafer than the cells we had bought in 2000.

Cypress acquired SunPower in three tranches. I had to fund the company personally because our board was tiring of acquisition “distractions” like SLM. It took me 15 months before I convinced our Board to invest in SunPower. Cypress taught SunPower how to make its world-class solar cell technology manufacturable by teaching them our professional R&D methods and how to manufacture at low cost. Key members of the SunPower manufacturing team came from Cypress’s lean and successful fab 2 operation. SunPower’s CEO and CFO also came from Cypress. Without putting trusted Cypress veterans in key positions, SunPower would never have achieved its phenomenal success. Cypress also transferred most of its business processes to SunPower, as well. SunPower uses our R&D process development system (NTP), new product development system (NPP), capital purchase equipment system (EPR) and P&L reporting method. Indeed, SunPower copied our business processes with such vigor that I noted they were in many cases better at following Cypress’s business process than Cypress was itself. They even located their first manufacturing plant near our assembly and test facility in Manila and were allowed to hire people from our plant (we have yet to recover from that loss of talent).

We took SunPower public at a valuation of $1.1 billion in Q4 2005. Although we maintained economic control with over 50% stock ownership, we also structured our shares to have seven votes each to give us over 90% of the voting power. Later, when SunPower offerings diluted our economic ownership to less than 50%, we still maintained control of the company, until we spun it out to our shareholders. Overall, Cypress invested $143 million in SunPower, and later sold SunPower stock for $677 million in cash and then distributed $2.52 billion in SunPower shares to Cypress shareholders in November 2008. The gain from our SunPower success is factors larger than the losses of all of our startup failures added together. In baseball terms, SLM, Ross and Multichip were only three strikes—one out—while SunPower was a game-winning, grand-slam home run.

The lessons learned from SunPower: 1) Use Cypress veterans in key positions. 2) To get star performance, put your star players on the team. 3) Transfer and demand the use of proven Cypress systems to startups. 4) Control shareholder voting with a super majority to prevent obvious decisions from being “democratized” and slowed down.

Silicon Magnetic Systems (SMS) was founded to bring to market the Magnetic Random Access Memory or MRAM, which at one time was assumed to be the “universal memory” that would replace all other memories—DRAM, SRAM and Flash—yet another Silicon Valley “hot technology” that did not live up to the hype. The SMS team included two Cypress veterans, Jeff Kaszubinski and Sam Geha, who were chartered to build a good startup culture and create bonds to Cypress. We needed to hire the key magnetics talent from the outside. That required separate equity, or “penny stock,” to attract employees, hence the need for a startup.

The startup team worked extraordinarily hard but could never get the MRAM technology to full production status. We eventually achieved economically viable 50% yield on a 256K MRAM but could never solve a nagging “blinking bit” reliability problem that was quantum mechanical in nature; i.e., a matter of science, not engineering. As we worked on solving the various problems that arose during the development of the MRAM, the solutions always added to the MRAM’s size and created more complexity. Eventually, the MRAM cell was twice as big as an SRAM cell (not 10x smaller, as hoped). Far from challenging SRAMs in speed, the 4ns write time of the magnetic material itself meant that we could make only medium-speed MRAMs that would never replace fast SRAMs. The final blow came when we added error correction to the MRAM to solve the “blinking bit” problem. That slowed down the MRAM by another factor of two—and it still wasn’t reliable. That was the end of SMS. The people who ran the project did the very best they could have, as I saw it, and have moved upward in their careers at Cypress. The project cost over $50 million before it was shut down.

The lessons learned from SMS: 1) Have the funeral earlier; we could have pulled the plug a year earlier and saved $15 million. 2) Use high-energy, high-integrity inside people to run ventures if at all possible. 3) Keep re-evaluating your business plan, especially as some of its basic assumptions become invalid—it’s too easy to keep “running on the hamster wheel.” 4) One of the strengths of Silicon Valley and Cypress is to allow managers to fail and then move on to bigger roles, as did both Kaszubinski and Geha.

Cypress’s latest startups are San Jose-based Cypress Envirosystems, which makes small systems to save energy, and San Diego-based AgigA Tech, which makes large non-volatile memories—another shot at doing what the MRAM was supposed to do. As of the writing of this memo, both companies are hopeful—but still losing money.


Cypress may create a startup to go into a new business, such as microcontrollers (CMS) or MRAMs (SMS), or it may be offered a technology such as BiCMOS from an external team (Aspen), or from an internal team, like PSoC (CMS), or it may want to enter into a new business for which it has no extant capability, such as microprocessors (Ross). In a typical case, the startup company is given its own separate equity, 7% to 20% of which (the common shares, also called “founders stock” or “penny stock”) is distributed among the founders and employees of the startup. Just as in the case of normal startups, Cypress invests in the new company by buying preferred stock, which has “liquidation preferences,” meaning that Cypress will own all of the startups’ assets in the case of failure. The original employees are sold common stock in their company, which is priced at a fraction, usually 10%, of the preferred share price—as justified by the liquidation preferences of the preferred stock.

By selling stock to employees, rather than granting options, the employees receive long-term capital gains treatment (about 20% tax), rather than the income tax treatment (35%-50% tax). Lawyers and accountants are quick to recommend granting stock options to employees, rather than selling common stock—because it’s easier for the lawyers and accountants. Explain to them once why you want to give out founders stock, and then, if you hear about stock options again, fire them.

Founder’s shares are technically called “restricted stock” because although the employee buys the stock, the company restricts the employees’ ownership of the stock for retention purposes. The company retains the right to buy back an employee’s stock if the employee leaves the company within a typical four-year vesting period (so-called reverse vesting). For example, if an employee leaves a company after one year, he or she can keep one-fourth of the stock and the company has the right to buy back three-fourths at the low price paid by the employee.

At launch, when the startup’s stock is priced at pennies, it’s no big deal for employees to write a check for a few hundred to a few thousand dollars to buy and own their own shares. Indeed, reluctance to do so on an employee’s part is a major warning sign. Later, as the common stock becomes more expensive, employees may not be able to afford to buy it. Only then should the company consider switching over to stock options, which require no cash outlay on the employee’s part but carry a heavier tax burden.

Nonetheless, having options taxed at a rate of 50% is such a huge penalty that startup companies often keep their restricted share programs going by lending money to their employees to buy their restricted stock outright. We did this—granting several million dollars in employee loans—at SMS. The inherent problem is that if the company goes under, the employees still owe the money to the company. To forgive the loans, which Cypress did in the case of SMS, leads to a big (non-cash) loss, even though no such loss would have occurred if stock options rather than restricted stock had been issued. This is the one penalty for pursuing restricted stock too long in the life cycle of a startup.

The initial valuation of the startup company (defined here as the preferred share price times the total share count of preferred plus common shares, both outstanding and ungranted in pools) is set to be a reasonable market value for the company as determined by comparables. For example, when Cypress Semiconductor consisted of nothing more than a business plan and six founders at our first round of funding, we sold 72% of the company for $7.5 million, implying a post-money valuation $10.5 million and a pre-money valuation of $3.0 million. Note that in the calculation above, the preferred and common shares are valued equally because in an IPO, the preferred shareholders exchange their preferred stock for common stock, which is a typical requirement of an IPO (new investors do not want a separate class of investors with superior rights).


Cypress Envirosystems was founded in 2007 to design Cypress chips into small systems and sell them for a multiple of the stand-alone chip value. Cypress Envirosystems is now nine quarters old, currently three quarters behind its revenue plan, but ahead of its cash flow plan, and still on track—I hope—to make its plan and be acquired in Q3 2010. The CEO, Harry Sim, is a Honeywell veteran who is smart, has good business sense and adheres to Cypress values about honesty. We have transferred significant Cypress talent (Bien Irace, Marcus Kramer, etc.) to help Cypress Envirosystems become successful.

I will use the attached Cypress Envirosystems one-page business plan as an example throughout this memo. You should detach it for reference. The initial valuation of Cypress Envirosystems, with all shares valued at the preferred share price, was $1.52 million (see the handwritten footnote “1” on the attached business plan labeled “A”). An alternate estimated market capitalization value for Cypress Systems in its first quarter of operation is $1.0 million, based on comparable companies in their seed funding stage (see footnote 2). The original investment of $918K to fund the company for its first quarter (3) was secured by selling 9.178 million preferred shares (4) to Cypress at $0.10 each (4). With the 6.00 million common shares (5) awarded to the founding team in Q1 2007, there were a total of 15.2 million shares of stock at that time worth $1.52 million, when valued at the preferred share price of $0.10. Thus, after the seed funding, the employees owned 39.5% of the company (6a), but that ownership was later to be diluted by plan to 14.9% (6b), as preferred shares were sold to Cypress on a quarterly basis to fund the company. If Cypress Systems employees find a way to fund their company without selling more preferred shares to Cypress, they will own 39.5% of the shares when Cypress Systems is acquired by Cypress in the future. Not to sell any more than the original 9.178 million shares would more than double the employees’ reward (due to reduced dilution and higher eventual share price). In a real startup, cash flow and equity dilution matter—unlike in the bonus plans used by big corporations. Furthermore, the original “deal” struck by investors and a startup is permanent. Stock changes hands. Lawyers are involved. Boards pass resolutions. There is no going back to recalculate or “reset.”

The initial equity picture of Cypress Systems seems reasonable: Preferred shares were sold at $0.10 to fund the company, common shares were sold to employees at $0.01, and we ended with a 60%-40% Cypress-employee ownership ratio. But where did those initial numbers come from, and how do we know they will make sense in the future? In order to answer those questions in a consistent way, we must first understand the P&L and cash flow of Cypress Envirosystems.


The first question that must be answered: “Is Cypress Envirosystems worth investing in?” Today, any BU that achieves $40 million in revenue with a 20% growth rate and 20% pretax profit makes a meaningful contribution to Cypress’s market value. The one-page plan for Cypress Envirosystems says (7) that the company will achieve this objective in Q3 2010, making its market capitalization about $100 million (8), which adds directly to Cypress market capitalization, using Cypress’s traditional 2.5x price-to-sales (PS) ratio (defined here as market capitalization divided by annual sales, calculated as fully diluted shares times share price divided by four times the prior quarter’s revenue). With about 165 million fully diluted shares, that means Cypress Systems will contribute $100 million or $0.61 to our share price in Q3 2010 if it achieves its plan. That’s enough to get Cypress’s attention.

The second big question is “How much investment does it take to achieve that result?” This question is answered by creating a standard P&L statement (9), as well as a cash flow statement in which, the free cash flow (10) is displayed along with the P&L in the one-page plan (A). “Free cash flow” is the operating cash flow (cash coming from or going into operations), minus investments in capital equipment; i.e., what’s required to fund the company.

The startup’s P&L and free cash flow statements should be created as philosophically faithful to the plan of a true startup as possible. For example, the startup company should pay its own employees from its own bank account, pay Cypress for rent and utilities if it’s on a Cypress site (11) and in general, pay for its own way from its own bank account. Letting Cypress accountants keep the books and make payroll can destroy the financial independence of a startup. When the startup keeps its own bank account and books, there is no charge for corporate SG&A (12)—true startups don’t have or want to have corporate overhead. Cypress Envirosystems has its own sales force. If it did not, there would be a commission for the use of the Cypress sales force (13) that would be negotiated at market rates.

Note that the planned net income of Cypress Envirosystems starts out with an $875K loss per quarter and stays negative for nine quarters (14). All of those losses need to be funded. In addition, there are extra cash flow needs to buy capital equipment and to fund “working capital” (inventory and accounts receivable). In its first quarter, Cypress Systems’ losses account for almost all of its negative cash flow (15), but after that (16), cash flow is a few hundred thousand dollars per quarter worse than the P&L losses. Note that the cash flow losses continue for two full quarters after the P&L statement shows a profit in Q2 2009 (17), necessitating $2.5 million in funding after the company becomes profitable. Cypress Envirosystems does not plan to become cash flow positive (18) until its third quarter of profitability.

In Cypress’s startup funding system, quarterly equity funding (3) offsets the cash flow losses, dollar for dollar. This mechanism forces a company to fund itself by selling preferred stock every quarter, which, in turn, requires the company and its investors to agree upon the company’s valuation (more later). The process of quarterly funding—having to raise money, knowing how much your company is worth, getting investors to agree—is the most important feature of the Cypress startup methodology. While we have copied the venture funding methodology as faithfully as possible in all other aspects, I believe that in this respect we have improved upon that system, at least for intrepreneurial startups.

The cumulative equity funding (19) represents the total amount of money invested to feed the cash flow needs of the company over time. When the company turns cash flow positive, the cumulative funding bottoms out at the value of the nominal funding plan for the company—in this case (20), $21.931 million for Cypress Systems. With this plan, we can quantitatively ask the question if the startup business plan represents a good investment for Cypress.

One way to look at the investment is the way venture capitalists do, by comparing the ratio of the market capitalization of the company created to the investment required. Depending upon the stage of the investment, venture capitalists might want a 5-to-1 to 10-to-1 return on their equity investment, as measured by the value of their stock at the time of a presumed IPO. As a company gets less risky and closer to an IPO, venture capitalists often bring in second and third round investors that view the company’s valuation in terms of a discount to the value that presumably would be achieved in an IPO. For example, just months before Cypress’s IPO at a price of $4.50 per share, we raised $10 million at $3.75 per share from so-called mezzanine investors who made a profit of $0.75/$3.75 = 20% in months.

In the case of Cypress Envirosystems, their business proposition is that if we invest $21.931 million (20), we will be delivered a company in four years that produces $9.99 million per quarter in revenue and $2.153 million per quarter in profit (7). As mentioned earlier, that would lead to an incremental $100 million valuation to Cypress (8), based on our typical PS ratio of 2.5. Thus, a $22 million investment gives a return of $100 million, a return on investment of 5-to-1, on the low end of the acceptable range.

One could make a similar calculation based on a nominal PE ratio of perhaps 12, which was typical or even conservative at the time of the writing of the Cypress Systems plan. The instantaneous market capitalization of Cypress Systems based on a PE ratio in Q3 2010 would then be annualized profit times the PE ratio: $2.153 * 4 * 12 = $103.3 million, essentially the same result as calculated by the PS ratio. (Note that taxes have been ignored in this simplified PE calculation.) Another way to crosscheck valuation result is to do a net present value (NPV) on the cash flow of the startup from the time it turns cash flow positive to estimate the cash payback time. Obviously, this calculation is subject to large errors, since it depends upon summing a profit stream that begins years in the future.

At this point, if the plan passes the simple ROI tests, it’s a go. If it does not, the team is either asked to go back to the drawing board to create a more aggressive plan, or the plan is rejected based on the belief that it cannot be made more profitable or less cash consumptive in a credible way. In the case of Cypress Envirosystems, we modified their first-pass plan by tuning down early revenue as being too optimistic and adding mid-term revenue from as yet unidentified products to create a viable four-year plan.


Now that viable P&L, cash flow and funding plans have been created; it is time to fit the employees of the new company into the picture. The approximate valuation calculations done above all use the pretax market capitalization of the whole company for making ROI calculations. However, employee equity has yet to be considered in the plan. Given the typical range of 7% to 20% employee ownership, a 5-to-1 pretax return to equity investors would thus be diluted to 4.65-to-1 to 4-to-1, respectively. If equity demands of Cypress Envirosystems’ employees amounted to 50% of the equity, the Cypress’s ROI would be reduced from the 5-to-1 enterprise return to 2.5-to-1 on our investment. In this case, the venture would be rejected as uneconomic. This situation often occurs during good times, such as 2000, when employee expectations are unrealistic (“We are the next Cisco; you just don’t see it yet”).

In the 1980s, we had a bias to launch multiple startups. After a good share of the early ones failed, as described earlier, I began to study the business processes of the venture capitalists that I deal with more carefully, both in their evaluation of startup proposals and as overseers on boards. The best venture capitalists reject 100 plans for every one they fund and do not fund companies that fail to meet venture capitalist Kevin Burns’ dictum that “Talent is King.” They know that it will be the “me-too” startups that fold first in every downturn.

Assuming that the demands of the startup team can be met with a reasonable equity budget (more later), the common stock equity plan of the startup must have certain relationships with those of Cypress. First, the value of the startup equity packages should be better than those from Cypress, so that the startup has preferential hiring power over the established company. “Preferential hiring power” means that the nominal capital gain of a restricted stock sold a new employee in the startup given at acquisition is significantly higher than the nominal value of the capital gain a new-hire employee would expect from his or her initial stock option or RSU grant at Cypress, as measured over a four or five year period. (For a nominal calculation, assume that Cypress stock will appreciate 10% per year over a four-year period, or 46% total, and make sure that the new employee receives an incentive1.5 to 3 times higher than that, assuming the startup succeeds.)

A better working definition for achieving “preferential hiring power” becomes available when the startup hires existing Cypress employees. As I outlined above, it is highly desirable for Cypress to put some star performers and key managers into each new startup. These Cypress top performers typically have many more options than other Cypress employees, requiring them to leave behind a significant amount of money to join the startup. In order to provide an economic incentive for a key employee to move from Cypress to the startup, anywhere from a 0% to 50% premium is needed. In some cases, Cypress employees are tired of their current Cypress job and will often swap an excellent equity position at Cypress for an equivalent equity position in the new company, especially if they are passionate about what the new company does. Once a few Cypress key employees are hired into the startup, future new-hire equity awards need to be based on comparables to the new mix of employees at the startup to maintain fairness among employees. This often will mean that the new venture must award more restricted stock to the best people it hired earlier from the outside (not just everybody) to bring them closer to the equity offers made later to Cypress transfers.

The discussion above very briefly presents the concepts used to create the Employee Equity Table, the fundamental document in the business plan that specifies the number of founders shares to be given to employees joining the startup. The equity package for the founders (for now, let’s assume they are all outsiders) are subject to negotiation in the environment of the current startup marketplace. Our startup CEOs have received equity packages that yield $1-$4 million at spin-in. The VPs of our startups have received $500,000 to $2 million at spin-in, with non-technical management rookies on the low end and seasoned Cypress technical managers on the high end.

The guidelines above are used to create the Employee Equity Table for the new venture, defined to be the stock value (not capital gain) that each employee in the company will have in the event that the company is spun in by Cypress. That list should have the following form:


# Employee Position Equity ($)
1 Name President 3,500,000
2 Name VP-Eng. 2,000,000
3 TBD VP-Sales 2,000,000
4 TBD VP-Marketing 2,000,000
5 Name Controller 500,000
6 Name Director-Design 1,000,000
7 Name Director-Systems 1,000,000
8 Name Sr. Des. Eng. 750,000
20 TBD Des. Eng. 2 400,000
21 TBD Des. Eng. 3 300,000
30 TBD Sales-Boston 300,000
50 TBD(6) Assy. Tech. 30,000 (6)

The list has several notable attributes. First of all, most of the senior employees of the new ventures (1-8) have the word “name” in the employee column, meaning that the founders and key employees of the new startup have been identified and an equity reward acceptable to them has been negotiated. Deeper in the list (20…), the technical ranks are filled out with “TBD” employees, whose reward has been predetermined in a consistent manner. Finally, the last-hired groups (30…, 50…), such as geographically distributed sales and manufacturing workers, are also accounted for as “TBD” hires. The Employee Equity Table contains the proposed org chart of the company at the time of acquisition. It also gives the sum of the equity awards is what Cypress will pay out at acquisition to employees. They will receive stock or cash for their vested shares and their unvested shares will be swapped for either Cypress options of equivalent value or for cash held in escrow until the vesting date. The early employees can consider the total share value on the chart above as the capital gain they will receive, given that the early common shares will be priced at pennies. Later employees will receive a lesser net award because their shares will be priced at a higher percentage (10%-50%) of the acquisition price. Thus, the new hire option plan need not offer fewer shares to employees in years two and three because a reasonable reduction of the reward for latecomers is taken care of by the common share price increase built into the plan. However, some of our startups have created plans that offer fewer new-hire shares to employees in the later years.

In the case of Cypress Envirosystems (see the “All-on-one-page” plan marked “B”), the total number of common share options offered to employees is 15.7 million (1). That equates to a Cypress payment to Cypress Envirosystems’ employees of $15.7 million at acquisition, which values the company at $1.00 per share for both preferred and common stock (2). The funds received from Cypress Envirosystems’ employees to by founders stock will have already been used by the startup to build the company and will therefore not offset Cypress’s purchase price as in the case of stock options. In effect, Cypress Semiconductor and Cypress Systems struck a deal that Cypress would pay $15.7 million to the employees of the new company to acquire their shares, given that the company achieved $10 million per quarter in revenue and 22% pre-tax profit (3).

The legal deal is different in several aspects from the business deal stated the last sentence. In reality, Cypress agreed to fund a company at the valuation levels outlined in plan B, which, at the time of the plan, represented the typical market valuations for a startup company of the type being funded. If the market changes, so will the valuations—they are not a promise; they represent a proforma plan. Cypress may also choose to discontinue funding the company at any time. We may do this because of a lack of performance or because we no longer want to be in the business we have funded. Cypress may also choose not to buy the company, even if it has achieved the revenue and profitability targets planned for the nominal acquisition quarter. Nonetheless, as rational investors, the construction of this plan will cause both Cypress and the company to proceed along the path of plan B (or mutually modified version of it). Cypress’s best interests will be served by acquiring an attractive new company in the case of success or recovering as much of its investment as it can in the case of failure. (More about the dynamics of acquisition later.) For the rest of this discussion, I will assume that Cypress and the company proceed along the nominal plan.


The equity plan for the startup can now be calculated in the nominal quarter of acquisition. For a company with $9.990 million in sales (3), a nominal PS ratio of 2.5 (4) and both common and preferred share prices of $1.00 per share (2), the market capitalization of the company is $100 million (5). Consequently, we now know that there must be 100 million shares (common plus preferred) outstanding in the acquisition quarter, when the preferred stock is converted to common and the company is acquired. If the employees hold 15.7 million shares (1), then there must be 84.3 million shares of preferred stock outstanding (6). The 15.7 million common shares come from the employee equity table. The preferred shares come from quarterly investments.

The question is whether we can create an equity plan with at-market funding for the startup that produces the desired end result. If so, the four-year plan can be constructed by connecting the initial plan (employee equity table, initial startup valuation, etc.) to the plan for the spin-in.

The 15.7 million common shares are well determined by the employee equity table described earlier. In the first quarter of the plan, 6 million shares have been earmarked for the early employees (6), the CEO and a few key engineers from inside Cypress. Given the uncertainty in the business plan and lack of a complete team, we agreed upon an initial valuation for the startup plan of $1.0 million (7), consistent with market values. This number could be much higher for a more established startup with existing products, such as SunPower, which was valued at about $20 million when Cypress first invested in it. In the case of Cypress Systems, the company needed $918,000 (8), which translated into selling Cypress 9.178 million shares (9) at an assumed price of $0.10 (10). This created an equity value of $978,000 (11) in the case that each class of equity was valued at its own price. From my perspective, the value of Cypress Systems at that time was $0.10 (the preferred share price) per share times 9.178 + 6.000 million total shares, or $1.52 million, as discussed earlier. Although the initial market cap estimation for the company of $1.00 million (7) is somewhat arbitrary, it must stand a market test in that a similar startup company funded at arms-length by venture capitalists or angel investors could be funded at an identical valuation without standing out as an anomalous investment. The Cypress Systems investment met this criterion. My guess (which can be confirmed by hiring outside valuation firms) is that any preferred share price between $0.05 and $0.20 would have been a reasonable at-market price to launch this company. Since the employees, nominal reward is determined by the employee equity table and not the initial valuation of the company, agreement on valuation is easily achieved.

When the company has little or no revenue, its market capitalization (7) is estimated on a quarterly basis by comparison to comparable external companies and to a venture investment database that Cypress keeps internally. Once the venture has achieved revenue, numerical market capitalization estimations such as the PS Ratio (ranging from PS=0.5 to PS=10, depending on the type of company) can be used to give more accurate at-market valuations. Once the company is profitable, both the PS and PE ratios can be used. The idea is to maintain the valuation of the company in the middle of the market at all times to allow us to keep on the proforma plan, assuming market fluctuations are not too disruptive.

While Cypress invests at a preferred share price of $0.10, the employees are allowed to invest at the lower common share price (11) of $0.01. This is the “penny stock” that allows the founding employees to buy their shares outright at a time when the valuation of the company is very low, so that they can realize capital gains tax treatment at the time they sell their shares to Cypress or to the market if the startup is taken public, as in the case of SunPower. When Cypress was founded, a typical ratio of preferred share price to common share price was 20-to-1. Today, a more conservative 10-to-1 preferred-common ratio is commonplace. The discount of the common stock relative to the preferred stock is justified by Cypress’s preference—the company has the right to keep all of the assets of the startup, leaving the common shareholders nothing, in the case that the company is liquidated. As the startup company successfully designs its first product, makes its first revenue, achieves breakeven profitability, etc.—the risk that the common shareholders will be wiped out by the preferred shareholders goes down. Hence, the preferred-common share price ratio must begin to move from 10-to-1 toward 1-to-1 at acquisition (2).

The creator of the plan now goes into a trial-and-error process to create an equity plan that is consistent with typical startup parameters at the beginning of the plan and with typical IPO or acquisition parameters at the end of the plan. The easiest first approximation is to insert straight-line preferred share price from the founding quarter (10) with a trial $0.10 share price, to the acquisition quarter (2), with a $1.00 share price. Likewise, a straight line can be used to connect the founding common share price of $0.10 (11) to the acquisition common share price of $1.00 (2). Since these two straight lines have different slopes, the preferred-common share price ratio will consequently be reduced on a quarterly basis. For example, in the first quarter of profitability, Q2 2009, when Cypress Envirosystems’ cash flow is still negative, but the company is hopefully on its way to success, the preferred-common share price ratio (12) is $0.533/$0.155 = 3.4-to-1. In the quarter before the proforma spin-in, the preferred-common ratio is only $0.887/0.710 = 1.25-to-1. In actual practice, the straight line approximation works for the preferred share price and curved line which keeps the common share price lower for a longer time works better for common share price, as can be seen by graphing the preferred and common share prices in the Cypress Envirosystems plan.

The preferred and common share prices are evaluated by the management team and the board on a quarterly basis. The purpose is to strike a business deal that both sides will accept, to make sure that the value of the company implied by the investment is at-market, and to reflect any significant changes in the market itself. In the early stages, this can be done by the company and its board, using information from other venture capital investments. In the last two years prior to the potential acquisition, the valuations should also be reviewed by an outside firm to certify that they are indeed at-market. There are horrible tax consequences (Reg. 409) for “undervaluing” the startup and giving its employees “cheap stock.”

The number of preferred shares sold each quarter is calculated by dividing the funding needed by the company by the agreed-upon share price. In actual practice, the preferred share price will deviate from that in the plan if the company is leading or lagging its plan. The amount of cash raised will also differ from plan. The common stock is issued to employees according to the employee equity table and may be given out faster or slower than plan, based on the rate of hiring. In the case of Cypress Envirosystems, all of the funding is assumed to come from preferred stock (a reasonable approximation) for two reasons: 1) the early common shares sell for pennies and therefore do not raise much money (14) and 2) when the common share pricing rises to a significant level (15), it is best to assume that stock options will be given out (no cash into the company) in the later rounds.

Thus, there is now enough information to fill out the preferred and common shares issued in every quarter to simultaneously satisfy funding and employee needs, respectively. The only plan variable not yet determined in the first approximation is the total number of preferred shares outstanding in the acquisition quarter (6). Of course, that number has to be empirically determined to make the Cypress Envirosystems plan self-consistent. In a typical case, the first straight-line preferred stock pricing effort will create too many or too few shares. This problem is addressed by adjusting the preferred share price curve in the first quarter. Note, for example, that if the initial preferred share price of 0.10 (10) had been increased to $0.12, it would have resulted in selling 7.650 million preferred shares in Q1 2007, rather than the 9.178 million shares in the plan. Thus, by adjusting the first-round pricing—minimally and within the at-market valuation limits—you can create more or less preferred stock—in this case, 1.528 million fewer shares in the very first quarter (with more savings thereafter, given the straight-line method used to calculate preferred share price).

This first-quarter adjustment is usually enough to create a straight-line preferred share-funding plan that is self-consistent with the rest of the plan, including the common share plan. Note that if the initial preferred share price must be made too high to achieve a self-consistency, the venture probably does not provide an at-market return to Cypress and the investment should probably be rejected. I am sure the case of an above-market return exists, but I have never seen one—Google never walked into Cypress to ask to be funded. The final check is to make sure that each of the common and preferred share price numbers make sense—to the investor, to the employee, to the IRS, to the SEC—in every quarter. (Here is where you will need legal advice.) If I were to critique the Cypress Envirosystems plan today, I would still find it very solid, two years later, but I would probably tweak the common share price of $0.71 in Q2 2010. It represents a discount of 20% relative to the $0.887 preferred share price, perhaps slightly aggressive for a company one quarter away from acquisition. Thus, the straight-line original preferred share price estimate should be considered as an approximation to be tailored—in the beginning quarters when valuation is highly subjective and in the later quarters when the company must worry about the tax laws surrounding cheap stock. If the quarter-by-quarter tuning of the preferred share price plan makes the plan inconsistent again, prorate the non-linear preferred share price curve up or down a few percent to re-balance the plan.

After combining employee common share ownership, preferred share funding, product milestones, the profit and loss statement, cash flow and enterprise market capitalization over a four-year period in a mathematically consistent way—while maintaining common business sense and meeting all legal requirements—the plan is now ready for approval. There is only one plan. This is it. It gets signed, board-approved and never changes. Once that first $918,000 check moved across to Cypress Envirosystems 9.178 million shares of preferred stock were issued to Cypress and the 6.00 million common shares were sold to Cypress Envirosystems employees, the deal was launched. There was no going back.

In practice, stock pricing tends to drift from plan. For example, Cypress MicroSystems succeeded in creating the market capitalization we planned on but the company was years late and spent $34 million more than the original plan to get the job done. That meant it had to sell extra preferred shares to Cypress, diluting the share count well above the nominal plan value. Similarly, if Cypress Envirosystems has to sell 104.3 million preferred shares total in the future, rather than the planed 84.3 million (6), it will end up with about 120 million total outstanding shares, rather than the planned 100 million shares. That would mean its planned market cap of $100 million (5) would be divided by 120 million shares to yield an acquisition price of $0.833 per share, rather than $1.00. Conversely, if the company can scrimp on cash or get customers to pay in advance or take any other measure to reduce the funding required by Cypress, the spin-in acquisition price of their shares will be higher than $1.00 per share. The internal startups that were bought out by Cypress, achieved acquisition prices ranging from $0.20 to $1.10 per share.

Once a startup has revenue, an at-market PS ratio can be used to help ensure proper valuation. Before that, in the venture stage, milestones (16) are explicitly added to the one-page plan to allow for objective valuation checks during the first year. Typically, milestones to hire top-quality key people are listed, along with creating the specification (NPP) to launch the first product. After that, we can track further hiring progress and follow-on design milestones, such as sampling new products and first revenue. The milestone section in the Cypress Envirosystems plan (16) includes milestones for two business units and placeholder milestones for “Concepts 4/5” to show when the second tranche of projects must be launched to fill in revenue in the middle part of the business plan.

The one-page plan is now complete. It gives revenue, gross margin, allowable R&D and SG&A spending, profit, free cash flow, required funding, cumulative funding, preferred share count and price, common share count and price, and proforma at-market corporate valuation, headcount, and major early development and hiring milestones—all in a self-consistent one-page format. If a new product does not sample on time, more R&D money is spent than planned to produce it, revenue and profit are delayed, causing larger than anticipated losses, increased funding requirements, and higher dilution due to above-plan preferred share sales. The final result is a lower share price at acquisition, which nevertheless only occurs when the revenue and profitability goals are finally achieved.

In the column describing the Cypress Envirosystems plan for Q3 2007, there are 36 numbers. They can be thought of as a vector that describes the company at a given stage of development. There are 16 quarterly vectors in the one-page plan that allows the board to accurately and quantitatively evaluate Cypress Envirosystems’ progress. In the early stages, cost control and corporate milestones dominate the plan and therefore corporate valuation and share prices. Later, the achievement of first revenue and control of expenses are most important. A year prior to acquisition, breaking even—in both profitability and cash flow—are key milestones. Finally, truly achieving 20% profit at a $40-million annualized revenue rate is the most important milestone of all in that it triggers the acquisition process (more later).