Strategic Business Development Jeffrey Byrne 1 Cyrix Situation

Strategic Business Development Jeffrey Byrne 1

Cyrix Situation: • Cyrix was selling a clock-doubled (core/bus) product at 80/40 Mhz. • Market memory pricing was outrageously expensive above 40 Mz • Market was ‘speed sells’ • Follow-on product promised to pick up the gap. I had serious concerns it would be significantly late. Actions: • Investigated clock tripled solution • Investigated ability to do it in ‘metal-only’ fix Results • Follow-on promised product was 9 months late • Metal-fix only solution worked and was quickly ramped in production. • Company stock price was saved, estimated hundreds of millions of dollars. • Generation of 3 quarters worth of product revenue (our only product) yielded 10’s of millions of dollars. 2

Cadence: TI Situation: • Start of services offering for Cadence’s Tality consulting services • Assigned on the account of Texas Instruments for worldwide services offering • Quotas set high for both software and services ($30 M/$12 M) • Services run rate approximately $2 M-$3 M/yr, misc AE on-site work • TI, last-of-the-dinosaurs, i. e. semi-internal offering (IBM, LSI, TI, Intel, etc) Actions: • Analyzed TI’s business. Talked to eng, exec’s, print, online, prior sales • Looked at macro-trend of business. Internal self-developed tools are expensive to do, difficult to hire for. Cadence/Synopsys/Mentor/other was EDA side. TSMC, Chartered, UMC, fabless model was the trend. • Broke business down into four major sectors/needs • ASIC – ASIC model such as National, TI, Fairchild, LSI, etc. – 3 rd party tools, timing/functional models, fault models, pattern generation, etc. – Library development, layout • Microprocessor effort • Analog/Mixed Signal • DSP Results • Worked with small TI team to propose changing the business model of yearly sales • Why keep selling the same thing every year, focus on growing business (revenue vs bookings) • Closed $100 M 3 year software deal. Tied up sales team, Blocked competition, drove revenue • Sold $11 M software consulting services for ASIC group to pick-up, move, translate from proprietary S/W to industry accepted 3 rd party S/W • Sold consulting services combination of design and S/W development to microprocessor group, mixed signal, and DSP groups. • Closed out services bookings to >$17 M, 160% of quota and 7 X increase in prior year results. 3

Cadence: Wireless/Wired IP Evaluation Situation: • Cadence had already organized ~7 separate BUs • Analog/Mixed Signal (Westinghouse team) • Datacom/Telecom (Nortel Ottawa) • … Digital IC ( U. S. , Livingston, Scotland) • Delivery targeted at ~85% utilization • Sales team selling well, but one-offs, hard to scale/leverage services • Desired to leverage that 15%, strategic platforms • Already decided on ARM as processor, DSP Group as DSP. • $ volume Product is with MIPS (Royalty) • $ volume services is with ARM (Services) Actions: • Often our IC solutions were bridging solutions. • Theory was to build something that is 80% what a customer needs • Value is: • TTM • Risky IP blocks already proven in silicon • Common bridges/platform already proven. -> $, margins • Had SERDES IP already, Evaluated wireless and wired technologies such as 802. 11, Home. RF, Bluetooth, HPNA, and Powerline. Results • Ruled out Powerline for various reasons. Ruled out Home. RF, Ruled out HPNA. • Bluetooth S/W stack was already developed by COMS group. Ideal for low power (i. e. custom silicon) 4 implementation • Picked 802. 11 for various reasons (prior to it’s huge success). Factors were market momentum, range, cost, power over Home. RF (Intel) and other • Worked on building a common platform with a wired/wireless pre-tested bridging application

Cadence: 3 rd party fabless Semi model Situation: • Cadence has an Digital IC business unit. Skillsets were front-back-end design • Selling engineering to engineering companies is not easy. What is value proposition? • Supplementing existing resources – Design center – On-site • New IP/Knowledge that others don’t have • Low risk, faster TTM, lower cost due to S/W tools and ‘rent’ model. • Skillsets we provided were not a focused core competency of the client. We were an enabler…. Actions: • Focused on low risk and TTM • Picked foundries that already licensed ARM/DSP Group IP and were leading 3 rd party foundries • TSMC • UMC • Chartered • Did test chips on risky IP • Developed SOC platform with bridging IP (see other SAR) Results • Successful in offering. One of only two profitable BUs (Dig. IC and Analog/Mixed Signal) • Of $300 M/yr, $150 M/yr was in hardware services • Of the 7 BUs, Dig. IC BU was $50 M/yr run rate, 1/3 of $150 M hardware services. 5

Lattice: Enabling Multiple Sales Channels Situation: • Lattice was developing/acquiring (Agere, other) complex FPGA solutions to go after the high-end FPGA market • Lattice Field Sales channel is composed of Lattice Sales Eng, Lattice Sales FAE, multiple Reps, and Avnet/Arrow for Distribution. • The programmable market is the inverse of a traditional high-volume semi market. Composed of 1000’s of customers and 10’s of thousands of opportunities. Need to leverage channels outside of Lattice’s dedicated North American ~ 55 field sales resources. • Need to create a value proposition against the market leaders of Altera and Xilinx • Need to communicate, educate the sales channel Actions: • Develop training for the sales channel • Technology-centric field training • Business/market-centric field training • Hold Distribution accountable for results • Created multi-module on-line study-as-you-go training (13 modules) with certification testing • Create KPIs and work with Distribution Mgmt to improve/get-rid of underperforming resources. • Pump up Disti mgmt on Lattice potential for them to make money. Results • I budgeted ($120 K) twice-yearly corporate training (2 -3 days) along with webcasts of targeted market 6 applications and block diagrams. Lattice CEO directed me to do market-training of Lattice Field Sales Engineers. • Results were ~103% growth Yo. Y on revenue for complex programmables • Avnet/Arrow results were increases of 30% Yo. Y on total socket wins with select (i. e. FPGA) socket increases of 108% Yo. Y. • Got rid of/replaced underperforming Disti resources, more of a burden than an asset.

Lattice: Opportunistic Exploit of SMPTE Situation: • Programmable offerings are heavily dependent on pins, memory, logic, power consumption, and SERDES. • SERDES are high-speed serial interfaces with embedded clocks, and run in the multi-GHz frequency range. Typically they are differential signals and drive cable, backplane traces, wire, etc. • These interfaces to backplane, cable, etc often are referred to as Phy’s for Physical Interface • Having an embedded clock greatly reduces board layout space and complexity although SERDES are extremely difficult to implement. • Lattice target markets were pretty broad, albeit Ethernet transport was common and the SERDES were designed to support Ethernet and SONET specs. • SMPTE (uncompressed high def video) used in broadcasting was very lucrative ($5/Phy channel) and had large customizable needs. An opportunity existed to exploit this tangential market using our existing offering with little investment. Actions: • Lobbied mgmt to obtain small funds to do an evaluation board and a few fundamental IPS (of which were already bought off on for std video segment) • Created marketing ppts on typical customer block diagrams of competitors solution, our solution and typical cost savings. • Created target customer list (~ 20) that could use this technology • Educated selected sales force (partial) where these customers existed and made joint calls with them Results • Technology was received well. Won a few tier 1 accounts for multi-channel video switch. Revenue 7 results unclear due to opportunity maturation from DI to DW and eventual Lattice cuts of 50% of workforce. • Increased pull-through of Lattice products due to my insistence on including our programmable clokk generator as a refclk source for SERDES. • Opportunity provided great insight into next-gen SERDES architecture (Quad, clocking, dividers) for SMPTE and traditional video segment.

Lattice: Narrowing Product Offering Situation: • Lattice had two primary FPGA offerings, a low cost, lower performance device and the acquired (but dated) high-end, high performance offering. Work was ongoing for next-gen in both segments. • Lattice had 2% market share compared to Xilinx and Altera. Most tier 1 default customers would design with the biggest, baddest FPGA out there as either a prelude to ASIC/SOC, or board componentry simplification. • A decision had to be made on where to make investments, which as a percentage of revenue we could not compete. Market/shareholder pressures to return to profitability, • Lattice fab partner was Fujitsu, being used previously from their embedded Flash technology. Fujitsu was using 40 nm tech and it was being used for volatile and non-volatile technology. Actions: • . Fujitsu SRAM technology was not keeping up with the industry (i. e. TSMC) and they chose to not invest in furthering the effort to smaller geometries. • High-end devices to be competitive would require next-gen technologies • Switching foundries would set back efforts 2+ years. • High-end devices bring high-margins, are great bragging rights, but may but bring in the large revenues for profitability (i. e. Qualcomm) • Choice as made to exploit the low-cost, low-power and non-volatile offerings of Lattice, as it was paired nicely with Fujitsu’s technology. • Bring a non-volatile instant-on FPGA to market • Develop/round-out a low-cost, low-power FPGA offering (ECP line) – Target low-cost, mid-volume, high-speed serial applications Results • High-end product development was cancelled. All energy went into the ECP l(volatile) and XPGA (nonvolatile) product lines. • Materials/training targeted Xilinx and Alteras lower-performance offering. Lattice was winning against these offerings. Socket wins increased 103% Yo. Y compounded for the Americas. 8

Lattice: Narrowing Market Focus Situation: • Lattice made some great strides to slim their costs to develop product by eliminating their high-end offering (see prior SAR) however this was still not enough. • Programmables by design are very flexible and can be used in a wide variety of applications including verticals such as compute, storage, automotive, consumer, video, mil, industrial, telecom, datacom, and other. Developing IP is very costly and having a 3 rd-best solution doesn’t win any business. • Lattice IP and applications teams were outspent by 10 -1. Next-gen technologies were often developed on an Altera or Xilinx platform. In order to win sockets alter, Lattice would have to provide significant value to overcome the risk to port. • The solution most evident was to narrow vertical focus and come out with attractive solutions in targeted verticals that were so powerful that the customer designed using Lattice from the beginning. • In order to do this, a narrowing of verticals must be accomplished. • A much better job must be done with literature and evaluation boards. Instead of focusing on feature validation, they needed to focus on solution validation. Actions: • A majority of FPGA business was being driven by the telecom/datacom segment. Growth in this segment was in the metro space, specifically metro-ethernet as QOS was improved. • Telecom/datacom business was being driven by greater needs for bandwidth. We believed this bandwidth was/will be driven by video • Lattice started designing IP and SERDES to provide fundamental video offerings with their low-cost, low-power solutions. • IP cameras were ideal as they could be updated to support next-gen video yet support legacy. Often transcoding was required and this used DSPs. • Low-power is important on battery solutions or POE solutions. • Additional growth was in wireless, the basestation market buildout. Pico and Femto stations. • Lattice targeted solutions for high-speed serial in portions of the basestation market. Specifically linecards versus core switching or processing. Results (2 x socket wins Yo. Y) • Basestation -> success • Video: too early to tell. 9 • Company after I left went after Silicon Blue ($60 M) for low-power ASIC-ish programmable mobile solutions.

Lattice: New product to replace aging CPLD Situation: • Lattice was known for being a main player in the CPLD market. CPLD solutions were typically viewed as ‘glue-logic’. Last minute fixes of super-fast logic, I/O voltage translations, and heavily pin-centric. Another characteristic is that CPLDs are instant-on, having embedded flash to configure the device, often used to bring up a board or a system. • CPLD fabric is physically pretty large, i. e. fairly expensive compared to other technologies. • CPLD fabric was sum-of products, limited levels, predictable timing, fast. • Parts were relatively simple, yet in the past had been very profitable. • I noticed during my field calls that the industry seemed to be moving from fast, predictable timing to more capacity-centric. Technology advances had made the logic was fast-enough and valuation was migrating to more logic, the need for memory, large pin counts, and more flexibility. Actions: • I lobbied my CTO and the VP of Major Accounts/New products to throw away the tried-true CPLD fabric has it has outlived its usefulness. Parts were fast enough and what was needed was more logic and memory , basically an FPGA-like part that is non-volatile and would replace the traditional CPLD. • It took 6 months to convince mgmt of the idea. I proposed leveraging our existing FPGA fabric to save engineering and bolt-on some Flash for configuration to provide the instant-on CPLD characteristic. Results: • The Mach. XO quickly took the lead in CPLD-ish sockets and wound up generating an estimated 40% of Lattice’s revenue moving forward. That revenue is estimated at greater than ~$65 M • Lattice had continued to add derivatives of this product to their offering. • Competitors (Altera specifically) felt threatened and created targeted campaigns to try to stem the success. of this product. Xilinx decided not to pursue this market segment. 1

Milestone: Re-segment market & Q/A issues Situation: • Milestone is the #1 provider of VMS software in the world. This VMS software supports 1000’s of streams of HD video for capture on a surveillance recording server. • Milestone had tow primarily offerings, a low/mid-end, low cost product and an enterprise product. • Milestones belief was the market being segmented into 3 levels of complexity, a low, mid, and a high-end segment. Their market strategy was to drive the mid segment, go after the low/mid segment and provide a path for growth/upsell. • In the U. S. , we were starting to lose our large customers as our S/W claimed 1000’s to 10, 000 streams per server and we were seeing failures of systems in the 400+ stream range. Actions: • Three primary objectives were clear to me these were: 1. The three tiered model was flawed. The lowest level was being encroached by the 2. 3. Results: Chinese, selling cameras and giving away S/W. The competitors were pushing their largescale solutions down to mid-complexity systems. We needed to go after the large-scale to be successful. The S/W, although being advertised/tested to go after large installs was seeing serious problems in the field in the U. S. I needed to convince corporate to fix the scalability issue. I needed to get N. A. onboard for focus on either scalability issues or new features. We needed to pick one or the other due to resource issue • I convinced my VP of the Business Unit and BD team that we needed to pursue the large- 1 scale installs/market. • I convinced engineering and the VP of the BU that we were seeing real problems (Milestone issues, not ‘customer issues’) and that we MUST do something about it to stem our already lost customers and starting-to-see-more of the same. • The engineering team was re-focused to provide two primary objectives. Scalability and migrateability of their products.

Milestone: Unique Value Prop for Server Situation: • Milestone is the #1 provider of VMS software in the world. This VMS software supports 1000’s of streams of HD video for capture on a surveillance recording server. • Milestone developed their own video server. This video server’s primary objective is to record large quantities of video 24/7. • Milestone’s server was designed using Intel i 7 mobile technology, competing against traditional Xeon-class general-purpose servers. • Milestone’s differentiated value proposition needs to be communicated to the sales channel to increase sales. Actions: • One of my first assignments was to drive the hardware server launch in the Americas. Upon reviewing the value proposition I identified two critical areas of improvement: 1. Video servers also provide value by only recording data on motion. This can be done by the camera detecting motion or the server. Milestones design (using Intel i 7) used dedicated hardware acceleration to provide server-based motion detection. All of the sales channel did not understand this technology and why it was better than the typically ITspec’d Xeon server. I needed to educate the sales channel 2. An Intel i 7 solution was really designed as a mobile platform, i. e. low power. Datacenters use a lot of electricity which is not green, and expensive. I reviewed the data of a competitors Xeon solution and Milestones i 7 solution with respect to TCO and cost to power the solution. I enhanced the value proposition. Results: • MARCOM materials were created showing the power of hardware acceleration and the field trained on this material. • MARCOM/training materials were updated to capture a new key value proposition for TCO. Based on a typical electricity rate, the Milestone solution (in addition to a better solution technically) paid for itself in 3 years based on savings of electricity only. 1

Io. T Market brainstorming n Io. T to me: an embedded application at an endpoint that provides value by collecting/using information from local/remote processing of analytics. u This has been around a long time. Enhancement in lower cost technologies and advancements on costeffective big-data collection and analytics are driving new possibilities Position/location-centric u GPS/GLONASS/Galeleo technology u MEMS/Dead-reckoning, accelerometers, etc. u Cellular triangulation u Stationary, location provided u IP address origination. Time Criticality Control Criticality Communication methodology: u Cellular u Store/Forward u Wired (Ethernet, USB) u Wireless (Zygby, BLE, BT, Wi-Fi, RF, NFC, etc) u Satellite u Real-time ( machinery & operations) u Batch (post-process, analytics) u Time-syncing (TDM) Target u Man-Machine or Machine to Machine? 1 u Machinery/operations) Data-centric u Machinery/operations) u i. e retail/transactional u Production u # units, hours, flow, etc. Interesting companies u GE- industrial/machinery u Honeywell/Siemens, others - Building/HVAC u Telit – hardware/chipset u Qualcomm – LTE, BT, GNSS u Bsquare – systems, integrator u KORE – worldwide, LTE-centric, one-stop

Knowles Unique Market Offering today u. MEMS • Microphone • Accelerometers • Other? Pressure sensors u. Audio algorithms • Noise processing/filtering • Speech recognition • Always listening u. Medical Industry Exposure? 1 Exit Strategy u. Sell/Acquire ? u. What part of the Io. T chain makes sense? • Who would partners be ideally? • What is the go-to-market channel? Future u. Audio + ? ? ? u. Medical/Sensor/MEMS? u. Industrial/MEMS? u. Where’s the beef? Now & future?