IP poker

Contracts, NDAs, indemnity and a host of other issues can stymie IP acquisition

By Bill Roberts, Illustration by www.mypetskeleton.com -- Electronic Business, 11/1/2000

Licensing third-party intellectual property (IP) cores and other virtual components is an increasingly important strategy for quickly developing systems on chips (SoCs), but getting the pieces licensed is a little like playing poker. As country singer Kenny Rogers advises: "You gotta know when to hold `em, know when to fold `em, know when to walk away, know when to run."

And know when to run like hell, says Carl Schlacte, CEO of Billions of Operations Per Second (BOPS) Inc., Mountain View, CA, which develops and licenses digital signal processor (DSP) IP cores: "Negotiations are high-stakes poker and brinkmanship. Ill behaviors leak out on both sides. Sometimes you reach a point where you understand the terms aren't healthy for your company. If the other guy gets everything and I get nothing, then I don't sign the deal."

Like most IP licensors, Schlacte can tell a few horror stories. But guess what, so can IP licensees. "We've encountered some IP providers who are unrealistic and price themselves right out of the market," says Mike Gulett, president of Virata Corp., Santa Clara, CA, which develops digital subscriber line (DSL) SoCs and software. "They wanted royalties that far exceeded the value of the IP to us. We walked away from those."

Licensees think time to market. Licensors think time to royalty. It is the same fundamental business proposition. Both have a lot riding on the expeditious transfer of IP cores. The trouble is, it's not often a smooth process. It's not nearly as straightforward, for example, as buying and selling chips.

"We've encountered some IP providers who are unrealistic and price themselves right out of the market." -Mike Gulett, president of Virata Corp.

An IP core deal can take anywhere from six weeks to two years to consummate. The licensee who does it in six weeks-a rarity-isn't going to miss the market window, leastwise not because of the IP core negotiations. The licensee who takes a year or longer-not as rare as the six-week-wonder-blows an opportunity for both licensee and licensor. "It can take six months to negotiate a deal, and sometimes the lifecycle of the product is only six months," says Steve Szirom, president of InsideChips.com, a division of HTE Research Inc., Bellingham, WA, who tracks the IP market. A six-month delay can knock half the profit out of some products.

The trend toward licensing IP cores is driven by the trend to build systems on chips, which comprise most or all of the functions that previously appeared on different chips on a board. With the number of transistors on chips now numbering in the millions, the only feasible way to fill them is with reusable sub-chip design pieces, many of them acquired from third parties. "To attempt to design one of these things from scratch within any market window would not be realistic," says Jim Tully, an analyst in market researcher Gartner Group Inc.'s London office. "So companies are forced to use these large reusable IP blocks and drop them into the chips they are making."

Neither buyers nor sellers are happy with the status quo. Licensees complain that licensors sometimes don't have the goods they advertise, make it difficult to evaluate cores and have grandiose notions of their IP's value. Licensors complain that licensees often don't know exactly what they need, want too much proprietary information before they pay and have unrealistic expectations for indemnification. Both sides complain about lawyers and long, drawn-out negotiations. In interviews, the word "lawyers" is often accompanied by nervous laughter or something unprintable.

The third-party IP market is still relatively small, but on its way to becoming a billion-dollar market (see bar chart). However, the buying and selling of IP cores is a business process full of conundrums and ambiguity in which, at the end of the day, licensee and licensor must both be willing to take a leap of faith.

Big buyers, small sellers

The technical issues alone are enormous. For example, there are no industry standards for IP core formats-and that's just one of a myriad of issues. The Virtual Socket Interface Alliance (VSIA), an industry consortium of users and providers, is defining standards, which it hopes will be widely adopted. Other organizations are working to solve business related issues (see sidebar).

The business issues stem from the characteristics of the market itself. Many third-party providers are small, while many buyers are big. There's no established business model for selling IP, which means revenue models are all over the map. Evaluating IP for the buyer's sake while protecting the seller's interests is an issue. Then there's those mind-numbing contracts in incomprehensible legalese with their indemnification and confidentiality clauses.

Rude awakenings are not uncommon. "When I first started to look at the IP market, I assumed it would follow the ground rules of any other market where you have a large number of small suppliers trying to sell to large buyers," says analyst Tully. "In those markets, the buyers dictate the terms. They say how much they'll pay and the sellers are grateful. In IP, it just doesn't work that way."

On the buyer's side, there's often a large chip maker or systems company with a time-to-market problem. On the vendor's side, there are a few modest-sized, well-known IP providers and dozens of shops with only a few engineers. Even the largest third-party provider, ARM Holdings Plc., Cambridge, England, isn't a large company. It only licenses about $90 million a year of IP cores, says Tully (see table).

Small providers tend to be technology driven, under-funded and sometimes mismanaged. "I've seen very small IP vendors dealing with very large semiconductor vendors who end up arguing, getting locked [in battle] on NDAs and contract terms. They get into...the kind of deal where the two legal departments go back and forth. It grinds to a complete halt," says Tully.

Risk and reward

One big issue is how to share the risk and reward. "The IP market is still very small and very new," says Tully. "There are no established business models yet. Even the top companies all have different business models. "

IP providers prefer up-front fees plus royalties down the road and, if applicable, fees for support during integration. Licensees will milk as much free support from providers as they can and would rather not pay royalties at all, although in many cases they must if they are to get the IP core they need. However, if the buyer pays too many royalties to too many IP providers there's no profit margin left. "If a company wound up paying royalties to 10 or 15 different IP providers, it would add up to more than the value of the chip," says Tully.

During the recent downcycle in the chip business, however, chip makers happily agreed to pay royalties because it was a way to defer payment, says Cary Ussery, CEO of Improv Systems Inc., Beverly, MA, which develops and licenses a configurable multi-processor SoC platform. "The royalty model allows two companies to share the risk and reward of using the IP," he says. "But people on the buy side are reluctant to accept it." Ussery says providers like Improv-whose cores are high-performance "star" IP-can realistically expect royalties, whereas providers of peripherals or libraries can't. "IP like ours is easier for the buyer to translate to their bottom line," he explains.

Even star IP's value varies from buyer to buyer and project to project. If an IP core is a key part of an SoC, then it's overall value is greater than if the same IP is a peripheral function on another SoC. "IP vendors must take a flexible approach to valuing their IP," says Tully.

Virata's Gullet says IP providers only harm themselves if they hold out for royalties when none are warranted based on the value of the IP to the chip under development. "Would he rather just leave the product sitting on the shelf or accept incremental revenue? A lot of providers haven't learned this lesson," he says. When Virata meets with such resistance, it will leave a function off an SoC rather than pay royalties, opting instead for a separate chip at board level to handle the function. "We don't want to pay royalties on every IP block because then we'd never get any money out of it," he adds.

On the other hand, licensees need to be careful not to squeeze IP providers too hard, cautions Tully. He knows of a case-and declines to name names-where a big customer squeezed a small IP vendor out of business, but didn't worry about it because it had written an escrow clause in the contract. "Then they ended up with the tapes of the source code and didn't know what to do with them," says Tully. "They devoted a lot of their own engineering resources to figure it out. It cost them much more than if they had just taken care of the vendor a little better."

"The IP market is still very small and very new. There are no established business models yet. Even the top companies all have different business models." -Jim Tully, an analyst with Gartner Group Inc.

A pig in a poke

The technology and business issues can't be entirely separated. One business issue is the technology evaluation process. Many customers take a lot of time evaluating IP before they license it, say IP providers. This requires IP providers to spend much time with the buyer, and sometimes to disclose more information than is comfortable. "If I'm the person taking the license, I want you to give me the IP and let me start my design while we're still negotiating," says BOPS' Schlacte. "As the licensor, I don't want to do that because it takes all the pressure off reaching a conclusion" in the negotiations.

Providers know licensing companies want to see benchmarks and numbers, but they say some buyers have the misunderstanding that they should see a finished product. "They want to buy the IP as if it were a chip, but soft IP is not a chip yet," says Improv's Ussery. "A lot of things can happen between the time they get it from us and implementation." Ussery wrestles all the time with how much evaluation support his company can give for free and when the customer must start to pay. There's no clean formula.

Extended evaluation periods prolong the acquisition process, benefiting no one if the end result is a missed market window. Ussery had one client that quickly knew it wanted Improv's technology, and hit it off immediately with Improv engineers. When talks started in earnest, the customer wanted more data, then wanted to run its own simulations, then took time to find the staff to run those simulations, and on and on. By the time they actually began to talk about a contract, six months had lapsed. It took another six months to do the contract, by which time a year had passed and a market window had closed. The customer did the deal, but didn't use the Improv IP until its next product.

The trick, from the provider's perspective, is to give the would-be licensee enough information to make the decision, but not enough so the customer can walk away and do the IP design itself-which has been known to happen. The industry is hampered by a lack of standard non-disclosure agreements (NDAs), says Andy Travers, CEO of the Virtual Component Exchange (VCX), Livingston, Scotland, a Web-based trading exchange for licensing IP. Even with an NDA, the great fear of every IP provider is that the customer's engineers will sign the NDA, get the component to evaluate, study it, decide not to buy it and then design it themselves. "There's a whole level of trust and agreement required between suppliers and developers that doesn't exist in the physical world," he says.

Buyers say who wouldn't be skeptical when considering an unknown provider peddling an untried IP core. "There is some technical due diligence you can do to evaluate the core," says Lou Lupin, senior vice president and proprietary rights counsel at the integrated circuit division of Qualcomm Corp., San Diego. "But a lot of time it turns on whether the core is well established in the market place or not." Qualcomm doesn't always walk away just because the provider is a small, less-known company. It does, however, lower expectations and routinely plan on using more of its own staff after it brings the untried IP in house, Lupin says.

Some buyers report downright bad experiences. "We got down the road with one IP vendor and realized they didn't have the ability to deliver the technology we were looking for," says Virata's Gullet. "It was a piece of IP not generally available in the market. So we had to find someone who could custom design what we needed. Fortunately, we decided this early, but it still cost us a couple of months in bringing the chip to market."

Some first-time providers have delivered defective IP, which Virata engineers had to work with to bring up to snuff, setting projects back from 30 to 45 days, Gulett estimates. Other buyers say some providers will lead them to believe they could have the IP at one price, only to try to extort more money after the buyer commits to using the IP.

"Buyers must beware," says analyst Szirom. "They have to understand the IP provider, know their capabilities, ask the right questions. If [the IP hasn't been put in] silicon [yet], they have to bear that in mind." Analyst Tully believes there's a budding business opportunity in this-the evaluation service provider. A neutral third party could build a business based around evaluating IP for others.

Companies also are experimenting with ways to speed IP evaluation by doing it on the Web. Simutech Inc., Portland, OR, has created eValab, a tool that allows testing of IP over the Internet. The product-at a price starting around $135,000-is slated for full production release this fall. "Nobody is satisfied with the way IP is sold today- evaluation of IP has been a major time and dollar sink for both IP manufacturers and buyers," said Doug Fairbairn, Simutech's CEO when he was interviewed for a profile on the company that ran in October issue of Electronic Business, page 52. Fairbairn is a past president of the VSIA, an industry consortium formed to expedite exchanges of IP (see sidebar).

	"The royalty model allows two companies to share the risk and reward of using the IP. But people on the buy side are reluctant to accept it." -Cary Ussery, CEO of Improv Systems Inc.

Contract woes

Buyers and sellers say they often can quickly agree to the general business terms of an IP deal and sum them up in two pages: pricing, terms of use, reuse, etc. It can then take months to resolve the contract.

"Liability, warranty, IP rights, co-development, disclosure issues, confidentiality, residual clauses-there are many legal issues," says David Almagor, CEO of MystiCom Ltd., Netanya, Israel, which develops and licenses DSP and mixed-signal IP cores. One problem is that the business terms are typically agreed to by a small group of people. The legal terms take a larger group. Unfortunately, Almagor says, the larger the licensee, the bigger the legal department, and the more eyes that have to see the contract.

Early on, IP contracts weren't that lengthy or onerous, but that didn't work either, says Timothy O'Donnell, the president of ARM. ARM did its first contract for microprocessor cores in 1992. "Some of our first contracts were 10 pages of legalese and two pages of deliverables," O'Donnell recalls. "I look at some of our first contracts and I shudder. That wasn't good for anyone."

By 1995, ARM had a standard, basic contract developed. Over the years, O'Donnell says, the number of legalese pages grew from 10 to 20 and the number of deliverable pages from 2 to 50. Today, ARM and other major IP providers work with relatively standard contracts. It helps that ARM and other major providers are well known. "I have seen contract negotiations take anywhere from six weeks to two years," says O'Donnell. "If it takes two years, you've killed any advantage to market."

"I have seen contract negotiations take anywhere from six weeks to two years. If it takes two years, you've killed any advantage to market." -Timothy O'Donnell, president of ARM Holdings Plc.

During contract deliberations, indemnity can be a hot topic. The licensee wants to be protected from any patent suits aimed at the IP. But indemnity can be expensive. It can also be useless, especially if the IP provider is a small company with few resources. If the buyer gets a solid indemnity clause and the small provider does get sued, it could put the provider out of business, creating a bigger problem for the licensee.

BOPS' Schlacte says: "If I'm free and loose with indemnity and give it to 20 or 30 customers, I'm vulnerable to 30 suits. If the indemnity cap is $10 per chip, and each licensee is producing a million chips, and I get sued, now I face $300 million in liability."

He says the tussle is between making sure the licensee is comfortable that the IP provider will stand beside the customer, but not putting the IP provider into the tank for doing so. Schlacte wishes the industry would come up with some kind of indemnity insurance. "Then it's just a matter of deciding who will pay the premium. When we have indemnity insurance, it will be a defining moment for the IP industry."

IP providers want to make sure any contract spells out precisely who gets to see the IP in the licensing company, how it will be used and how many times. And they want to know how the licensee will protect the confidentiality of the IP from other engineers within its own company and from its partners. "If I license a process to you for one product, what's to keep you from using it in another," says Schlacte. "We want to know to what degree our technology is fire-walled from the customer's other projects."

One way to speed up contract negotiations is to move on parallel tracks with business, evaluation and contract issues. One of Improv's best experiences came when a customer agreed to have the technology discussions, business discussions and legal discussions concurrently. It helped that the customer was a smaller company, had a clear product in mind with a specific and narrow market window, says Ussery.

Improv demonstrated the technology. The customer liked it. Improv sent its standard contract to the customer's lawyers to get started, and began to evaluate the IP with the customer for its particular use. "We did the deal in six weeks," he says. Ussery says the IP acquisition industry needs more examples like this. One reason it doesn't have them, he believes, is the legal department of a large acquiring company typically doesn't want to see a contract until all the technical, engineering and pricing bugs are worked out.

In us we must trust

No matter how much attention two parties pay to dotting all the "i's" and crossing all "t's" in the contract, at some point-if the partnership is to work--they must move ahead. The irony, says Schlacte is "you examine all the things you don't trust about each other, then you just go ahead and trust each other." If he really didn't trust the other party, he probably would have walked away long before he got to the contract negotiations, he says.

Some providers believe formulating long-term partnerships is the real key to success in the IP business. Schlacte typically won't do a license as a one-time deal. The first license always takes the most time, and subsequent deals with the same customer can be quickly reached, he says. This puts a premium on building relationships that will last and the customer will use the providers IP cores on future projects.

No one believes the IP core licensing model is going away. On the contrary, it is becoming ever more important to the industry. But because the model is still immature there have been, and will continue to be, problems. None of the people interviewed for this article are put off by their own occasional bad experiences, and they all have plenty of positive experiences to keep them in the game. Despite the fact that all of them had faced some challenges in the past, all of them are optimistic that the model can evolve into a smoother operating effort.

Going forward, the best chances for long-term success would appear to lie with those licensees and licensors who build close partnerships that will yield results across many projects over several years. To do so requires a lot of trust, and it takes time to build trust. "Trust is a big issue," says analyst Tully. "But somehow, trust has to play a much bigger part in these deals than it does now."

---

Bill Roberts is a contributing writer to Electronic Business. Reach him at brobert1@ix.netcom.com.