So, you've been walking, driving, even showering with this great idea in your head. The time is right to be your own boss and take your idea to market. But you're not sure what it will take to get there.
The experiences of Nazomi Communications, a fabless chip maker that invented a Java accelerator chip and brought it successfully from conception to launch, may serve as a case history.
The first task is to leverage what you know and are passionate about. Use your expertise. Nazomi's team used its extensive experience with systems semiconductors and Java and a deep knowledge of the markets to create a silicon-based Java acceleration solution.
Second, when talking to your target customers, don't be tempted to ask them what they're looking for in terms of product specifics. Their suggested solutions will be limited by their perception of the problem. Instead, determine the experience that is desired. Remember, you're the one with the expertise; it's up to you to come up with the solution that will enable the desired experience. For Nazomi, the challenge was to provide faster, more reliable Java software performance. Leveraging their expertise, the founders invented and patented a silicon-based acceleration solution.
Once you have solidified the solution that you will offer and have done a good amount of research to confirm that it has potential, it's time to develop your product. This is where the sweaty palms and palpitations start, because to develop a product, three things are needed: a good idea, a team that can execute and money.
The idea is already there, and you've established it's a good one. Check off that box on your "to do" list.
It's a good decision to be part of the development team for the product, since you can use your expertise and knowledge of the concept to lead the project. Look for like-minded professionals with the skills required to architect it. After you've found them, you can check off "design team" on your li st.
As for the money required to run the development and subsequent production, unless your Aunt Nelly left you a hefty inheritance, it'll have to come from venture capitalists or other sources.
How do you get into the wallet of a venture capitalist? There are a couple of methods. One is good, old-fashioned cold calling-over and over again-to get your foot in the door. Be persistent, and perfect your three-sentence introduction on why you should be heard. Alternatively, you can find someone with the proper connections who can introduce you. That method tends to speed things up considerably.
A tip from the founders at Nazomi: First,select one or two sources whom you don't really expect to come through with cash, and pitch them early in the process. That will give you some practice and allow you to perfect your pitch before you meet with the VC sources who really matter to you.
In your pitch, highlight not only your idea but als o the business it can be expected to generate and how you will go about achieving those business results. Be mentally prepared to be rejected a lot. (Remember trying to land a date for the prom? This is much worse.) But don't give up, and keep evaluating your pitch. Does the business plan make sense? Can it be tweaked and perfected? If it can, make sure you follow through.
Eventually, just like prom time, someone will say yes. Now the money's banked; time to get the product out.
By now, you have defined the many stages and segments involved in creating your product. Individual team members will work on different building blocks to shape the initial, larger design. Many times during the development process you will face the question: "Should we do this internally or farm it out?"
To determine the answer in each instance, calculate your return on investment. If you need to hire multiple people to complete a segment but are unlikely to require that expertise later on, you mig ht be better off outsourcing. On the other hand, if it would require specific and extensive training to understand the technology and the task at hand, in-house design may be the better option. That holds especially true if the knowledge gained from the design process remains valuable throughout the development of the product.
After the outlines of your design and the basics are defined, you will need extra manpower to fill in the details. For Nazomi, after the initial architecture was built, teams of hardware and software engineers were needed to write the specific architecture that would become the JA108 chip.
While the engineering team is working hard at developing the product, it is time to initiate your business development. Contact your target customers (those who seek the experience and results that your product can offer) and, under the necessary nondisclosure agreements, tell them about your solution and solicit their input. This is a good time to find out about features that you may not hav e considered but that would make a valuable addition to the product. It is also a good time to reconfirm the timing of your product: When will the customer need it? Will you be in time to accommodate that schedule?
It is always a good idea to have your technical literature ready and up to date when pitching your product. Samples will work even better.
Nazomi, in this process, discovered the ever-increasing need for Java acceleration in wireless devices, especially wireless handsets. Wireless carriers had started to look at Java-based applications as a means to grow their average revenue per unit. They needed fast, reliable Java phones to deploy their services. To accommodate that need, the handset OEMs were looking for the best Java acceleration, with the quickest time-to-market. Nazomi, with a chip that could increase Java performance up to 200x, found the time right and the target customers eager to learn more about its solution for fast and reliable Java performance.
Meanwhile, back at the off ice, the design of your product is nearing completion. The architecture is in place, and all segments of the design are finished. But before you can go into production, you must conduct extensive verification to assure that your product will be fully functional after fabrication. Nazomi, in testing the design for the JA108, undertook multiple levels of testing. That entailed a detailed understanding of the design and the modular interfaces in order to target functionality testing efficiently.
The first level was simulation and gate-level testing. After passing those tests, the JA108 was tested in a field-programmable gate array in a board-level environment. The goal of the FPGA implementation is to simulate many more cycles than are possible on a behavioral simulator. That allows for validation of the architecture and the design at the system level, including the design's ability to run with operating systems, associated drivers and related software stacks. Assurance of complete functionality at the syst em level, for hardware and software components, reduces time-to-market.
Testing is a time-consuming process. The best way to shorten it is to leverage tests already written for similar applications. Nazomi has successfully used FPGA and other methodologies in the development of its silicon intellectual property. Since the JA108 chip is based on that IP, past development efforts were fully leveraged, thereby shortening the test cycle.
At this point in the process, you can feel the excitement. You see the light at the end of the tunnel, and it's not an oncoming train. Your design is done; the testing tells you it's good to go. Now you must have a way to produce it. That brings a new set of questions: Who? What? Where? How? How much?
First, define the product specifications. Then list the sources that can accommodate those specs. Finally, identify which vendor offers you the highest ROI. Sometimes a turnkey solution is exactly what you need. Other times, you might want to t ake on some aspects in-house to stay within budget.
Let's look at Nazomi's decision process. First, the specifications. Nazomi's JA108 is a low-voltage, high-performance chip, which is ideally suited for a 0.18-micron process. In looking at the foundries that can accommodate that process, factors such as capacity, cost and fab availability were evaluated. Another factor was the availability of public design libraries, since Nazomi preferred to go that route rather than design its own libraries.
Nazomi also outsourced part of the chip development process. That process proceeds as follows: After testing confirms the functionality of the chip design, a number of steps must be executed before first silicon is produced in the foundry.
First is the synthesis of the Verilog coding (the schematic of the chip) into a gate-level netlist and the generation of pre-layout timing. The netlist then needs to be laid out, a process during which place and route are accomplished. At that time, all the layers are la id out. Post-layout parameters are extracted and fed back into a timing analyzer. Corrections are then made, and another round of layout and verification is done.
The chip is now ready for tapeout. The layout goes to the mask shop, where the different mask layers are made. The masks are then used to produce the wafers through etching and deposition.
After your product is fabricated, usually it will need some packaging. The criteria for selecting how to package your chip, and who should accomplish it, are pretty much the same as those for choosing a foundry: Determine who can accommodate your specifications, your budget and your timing better than anyone else. An important factor to consider here is that both freight and time are cost factors. For example, if both your foundry and customer were in Taiwan, having your chips packaged in the United States would cost you in lost time and freight expenses.
Back to our production process: Once you have your product in your hands, you will need to conduc t more tests to make sure it is functional and does all the things you claim it will, for as long as you claim it will, without blowing up or causing weird side effects to the user. When the JA108 arrived from the foundry, the stage was set to bring up the chip on an evaluation board and run applications and software. That allowed for testing of the functionality of the chip as well as for confirmation that all electrical specifications were met.
So, first came the idea, then the market research, a team to make it happen and the necessary funds to see the project through. You carefully planned and executed the development of the product as well as the business development. You tested and tested again. You made the right choices for production and packaging.
Now you are ready to conquer the world with the perfect product to provide your customers with the results they desire.
Nazomi did it; so can you. Go for it. It's a blast!
--- Mukesh Patel is president and chief executive officer of Naz omi Communications (Santa Clara, Calif.). Before founding Nazomi, he was senior manager of Sun Microsystems' Java and Ultrasparc microprocessor design group. Patel holds a BSEE from the University of Wales, U.K.
Copyright © 2002 CMP Media LLC
5/1/02, Issue # 14155, page 33.