It's about delegation

Why make the main CPU work hard, when you can have a much cheaper chip do the job?

In the physical world, when we send letters, someone comes, pick them up from the mailbox, and delivers them to the addresses. We’ve gotten into the habit of thinking that it works the same way with the Internet, but it doesn't. In order to streamline the transfer of information, every file sent by Internet is chopped into many small packets, each of which reaches its destination separately: one by way of Italy, another by way of Australia, and a third through Somalia. Thanks to an address on each packet, called a header, each is able to reach its target. The computer receives the e-mail, and puts the pieces back together in the right order. Doing that, however, requires a great deal of calculating power.

Tehuti Networks likens this final stage to the selection of mail reaching an office. “According to the conventional solution, the ‘manager’ the computer’s main CPU, does the job,” says Tehuti CTO Haim Bar-David. “Our goal is to assign a large part of the job to a special chip that does nothing else. This is much simpler and cheaper than having the CPU do it. We compare it to a secretary, who is hired to take as many routine and simple tasks as possible off the manager’s hands.”

”Globes”: Up until now, all the packets have been processed by the CPU?

Tehuti CEO Arie Brish: ”Not exactly. Many companies are trying to simplify the process. Some give the entire process to hardware, i.e. a separate chip, but then the chip is very expensive. We compare this solution to a very efficient and intelligent, but also very highly paid, secretary. Others attempt to do a lot with software, and only a little with hardware. We’ve found an algorithm that is the best combination of the two, and is therefore the cheapest. We’ve found a unique way of enabling the chip to get the most out of the computer’s memory, and that’s what we’ve patented. Since the patent, however, discloses a lot, and it can be bypassed, most of our information is not recorded in the patent. We’re simply making sure that it stays a secret. In order to make it even harder to solve, we don’t use the standard products on the market. We didn’t buy chips; we developed them ourselves.”

Links measured by thickness

Ben-David was a hardware engineer at Siliquent, where he specialized in communications protocols. He formerly worked for MMC, a start-up that was acquired by communications and data storage company Applied Micro Circuits Corp. (Nasdaq: AMCC), He felt, however, that things were moving too slowly there, and in an unclear direction, so he preferred founding his own company.

Brish, who is the man with connections, has an unusual record for someone in start-ups. He was a partner in initiating and founding many and diverse projects in Israel and the US over the past 25 years, including National Semiconductor (NYSE: NSM), Motorola Israel, Motorola’s DSP center in the US, Conformative Systems, VENewNet, and cxo360. In 2000, he began working as a consultant for a number of venture capital funds, where he came across Tehuti.

”The software processor market is a very young market, only 3-5 years old. All products are still prototypes, except for one that is very expensive. Our product is the only reasonable priced solution for home users.”

What do you consider expensive?

”A chip that costs $150 is expensive. It just doesn’t pay for a mobile or desktop computer user to buy it. That leaves us with the entire private market.”

Will you sell the chips directly to users?

”No, that’s not realistic. We’re acting on two levels. One is simply to produce chips and sell them to large companies through OEM agreements. Later, we plan to outsource computer components and cards, on the basis of our ability to distribute work optimally between hardware and software. We’re in touch with all the large computer companies: Dell, Hewlett-Packard, and IBM. The responses are very positive.”

How does a company your size even manage to get huge companies like IBM and HP to listen?

”In a start-up, personal connections are worth just as much as technology. If a large company is competing with us, it has 20 sales representatives who know HP. All it has to do to schedule a meeting with HP is ask the salesperson to call. Because I worked with Motorola, I can rely on salespeople there whom I know to do me a personal favor by making an effort for me.

”When a new company which has no such connections goes overseas, however, it has to find a company to represent it. Such a representative is valued by the thickness of its phone book. It is usually paid both a fixed fee and commissions, so that even a small company can afford the price.

”Even for a company like that, it’s important to develop personal connections. It takes more than charm; it takes a long process of accumulating information about the company to understand, for example, who really makes decisions, which department has more money to spend, which senior executive is more enthusiastic about new technologies, and who likes meeting Israelis. You can also learn this from junior employees in the company, or from suppliers and customers who work with it. Most of them are glad to help: they know that some day, they might need you.

”Some entrepreneurs rely on the connections of technology incubators. Incubators really do have connections, but that’s not their official job. The incubator is designed to provide administrative backing. You can’t rely solely on its connections.”

Ever-falling prices

Tehuti raised $1 million in its seed round from ProSeed Venture Capital Fund, Alice Labs, Lachman Goldman Ventures, and the Technion Research & Development Foundation. The current investors have also undertaken to participate in Tehuti’s upcoming round, in which the company plans to raise $6-7 million for development. Before it starts making sales, Tehuti will have to raise a further $15 million, but it already has a trial with one customer.

Do you have an agreement with a manufacturing facility?

”We already have a connection with a manufacturing facility, but there’s no point in signing with a plant before we develop a prototype. Prices are constantly falling. At the same time, we must sign a contract with the facility before we sign our first sales contract, so that we can commit ourselves to a price. As a matter of principle, it doesn’t pay at all to build a chip foundry in Israel. We’re planning to outsource our production to Taiwan.”

Where does the name Tehuti come from? It sounds like something from mythology.

”Tehuti is hieroglyphics. When the company began, it dealt with encoding for information security. The company has changed, but the name hasn’t.”

Published by Globes [online] - www.globes.co.il - on October 31, 2004

Twitter Facebook Linkedin RSS Newsletters âìåáñ Israel Business Conference 2018