* * *

• Upgrading BRCM from BUY to STRONG BUY given 1) we think the firm has won more Bluetooth designs for 2H with Nokia (NOK) than most anticipate, 2) we expect chip stocks to appreciate in 2007 with BRCM outperforming most peers given its diverse product offering and technology leadership, 3) we believe the firm’s very heavy investments in 65nm products are likely to pay off in 2008, and 4) new product launches at 3GSM increase the likelihood of a 2008 cellular product cycle with Nokia and/or Samsung.

Initiating coverage of Marvell with a HOLD rating and $20 target. We think MRVL is fully valued relative to its 2007 and 2008 earnings power and look for 1) a cheaper stock near the $15-16 range, 2) signs of greater than expected share gains in its handset chip business, or 3) greater than expected growth in Wi-Fi, Ethernet, or hard drive chips. We greatly respect Marvell’s technology portfolio and execution track record, and look forward to recommending the shares at more reasonable valuation levels.

Checks show near-term business trends remain choppy. Recent checks indicate that Marvell has not seen a ‘snap-back’ in its current business trends heading into the seasonally slower first quarter. Furthermore, checks indicate that Marvell continues to take steps to prevent its internal inventories from growing too large, another sign that demand has not materialized as expected.

Operating margins are significantly depressed due to recently acquired Intel (INTC) handset chip business. Marvell’s new handset chip business is driving its operating margins down from 28.6% in April’06 to 5-6% in April’07. In order to capitalize on its large and risky investment Marvell management must: 1) win 3G chip share in a very competitive market against TI, QUALCOMM (QCOM), Broadcom (BRCM), and others, 2) reduce chip production costs, and 3) scale operating expenses. Also, there is integration risk in that corporate cultures may not blend, management may be distracted from its core business, or technology integration may be problematic.

Top-line growth drivers include chips for handsets, embedded and 802.11n Wi-Fi, GbE infrastructure, and printers. Marvell’s largest growth opportunity is its baseband and applications processor segment (+20% YoY in 2007, acquired from Intel) as the firm tries to penetrate 3G markets and push its technology down into the mainstream handset models. We believe the baseband and applications processor chip markets total $12 billion per year, of which Marvell has 3-4% market share. The firm’s other 2007 growth opportunities include ramping embedded and 802.11n Wi-Fi chips (+46% YoY), driving Gigabit Ethernet chips deeper into the infrastructure market (+10% YoY Ethernet overall), and ramping printer chips (+59% YoY) for HP and other customers.

Core storage business (53% of revenues) is more mature with single-digit growth likely in coming years. We believe that Marvell’s hard disk drive storage chip business is a more mature and slower growing business these days as unit growth is largely offset by ASP declines. We see incremental customer share gain opportunities largely limited to Seagate, and incremental socket share gains limited to lower content motor controllers and pre-amplifiers. While Marvell has done a great job of winning market share over the past five plus years (39% in 2006 per our estimates), many of its growth opportunities are largely exhausted.

Marvell has significant competitive barriers to entry. Marvell has assembled a very broad suite of intellectual property in chip design, software, and reference platforms that drives significant barriers to entry. The firm leverages its broad suite of technology by integrating many different functional components onto a single chip, thus driving small solution footprints, low costs, high performance, and high functionality.

Valuation is rich at 29x forward EPS estimate and 4.1x EV/S multiple. Shares of MRVL are trading at somewhat rich valuations, in our view. Our price target is based on a 30x multiple of our forward earnings estimate (we are using 2H’07 and 1H’08 to account for the firm’s topline growth opportunities), slightly below the firm’s four-year historical median forward P/E multiple. Our price target is also based on a 4.2x enterprise value-to-sales multiple (2007), also below the firm’s four year historical median multiple to reflect Marvell’s depressed margin structure and likely slower growth rates going forward.

Risks to attainment of our share price target include: options backdating related risks, worse than expected chip demand, a slower than expected increase in handset chip operating margins, less than expected market share gains in handset or Ethernet markets, more than expected chip competition, and chip market cyclical downturns.

MRVL 1-yr chart

Revenues are likely to reaccelerate on a year-over-year basis in June 2007. We forecast Broadcom’s revenues will begin reaccelerating on a year-over-year basis in the June 2007 quarter, a positive given that chip stocks have historically trended in concert with year-on-year revenue shipments.

Broadcom has several high profile sockets that could help drive growth in 2007 including with Nintendo and possibly with Apple and Nokia. We believe Broadcom has several high profile sockets that could drive meaningful growth in 2007 including: 1) Selling Bluetooth chips into the Nintendo (NTDOY) Wii, 2) Possibly selling Bluetooth + FM Tuner chips into Nokia’s (NOK) handsets ($150 million opportunity), and 3) possibly selling mobile multimedia chips into the mythical Apple (AAPL) iPhone ($140 million opportunity).

Ramping product cycles remain intact, likely to drive future growth for chips including: 1) Bluetooth into handsets and headsets, 2) 3G handsets, 3) Wi-Fi 802.11n, 4) gigabit Ethernet into switches, 5) digital TV, 6) voice functionality into cable modems, 7) mobile multimedia chips, perhaps into the mythical Apple iPhone, and 7) other applications like VoIP processors, GPS, and power management.

Broadcom is also exposed to some more mature, slower growing markets. Broadcom does have about 60% exposure to some more mature, slower growing end markets, partially a result of its large share gains and revenue growth in recent years. We believe slower growing markets include chips for 1) the Ethernet market (client and infrastructure), 2) cable and satellite set top boxes, and 3) cable and DSL modems. These three major product areas, which comprise about 60% of total revenues, may continue to grow for Broadcom if the firm can still win meaningful share from competitors, or if the market research forecasts are too low.

Broad IP portfolio drives significant barriers to entry. Broadcom has assembled a very broad suite of intellectual property in chip design, software, and reference platforms that drives significant barriers to entry. The firm leverages its broad suite of technology by integrating many different functional components onto a single chip, thus driving small solution footprints, low costs, high performance, and high functionality.

Has the leading market share position in multiple end markets. Broadcom has the leading market share for chips into multiple end markets including cable modem, cable STB, satellite STB, GbE controller, PHY, Ethernet switches, Bluetooth, and Wi-Fi chips. Broadcom does not just want to compete, but rather dominate its chosen markets, and the firm pushes the integration and product diversity envelopes to achieve its objectives.

Valuation is generally reasonable at 22x forward EPS estimate and 4.4x EV/S multiple. Shares of BRCM are trading at reasonable valuations, in our view. Our price target is based on a 30x multiple of our forward earnings estimate (2H’07 and 1H’08), in line with the firm’s four-year historical median forward P/E multiple. Our price target is also based on a 6.1x enterprise value-to-sales multiple (2007), also in line with the firm’s four year historical median multiple. Shares are currently trading at a 30% discount to these historical median multiples.

Risks to attainment of our share price target include: options backdating related risks, litigation risks with QUALCOMM (QCOM), customer concentration risks, a cyclical or end demand slowdown in the semiconductor industry or macroeconomic environment, market share losses in key markets, technological difficulty in moving its chips aggressively to 65nm, worse than expected chip pricing pressure, or weaker than expected margins.

BRCM 1-yr chart

• Checks suggest IR’s commodity chip supply remains very tight in the distribution channel, with recent double-digit price increases on FETs and leadtimes lengthening to 20-26 weeks on some parts. Recent double-digit price increases and lengthening commodity lead-times (20-26 weeks on some parts) do reflect strong demand from ramping gaming consoles and new Intel server platforms.
• IR’s supply tightness reflects robust demand from ramping gaming console and Intel server platforms, but also reflects less than stellar planning or execution in ramping its new fabrication facility. IR’s supply tightness is a direct result of several factors including 1) strong demand from ramping gaming console and Intel server platforms, and 2) less than stellar planning or execution in ramping its new fabrication facility.
• While IR’s recent gross margin performance has been disappointing, we wonder how much of the June ’06 bad news was driven by product cost spending versus non-recurring other cost of sales associated with ramping the fab. IR’s gross margin performance has been disappointing for several quarters. Nevertheless, we wonder how much of the gross margin bad news was driven by fab expenses that should ramp down over the next quarter or so, like fab-line start-up charges or underutilization charges for its new Wales facility, versus how much spending is recurring COS.
• While we respect International Rectifier’s server and gaming console design wins, we remain at a HOLD given: 1) the firm is not making sufficient progress towards its 50% gross margin target, 2) IR seems to have made material margin concessions on its new gaming and server platform wins, 3) the firm is ramping up its OpEx spending significantly, and 4) the stock’s valuation multiples are not compelling relative to peers or its growth rate. While some momentum players are likely interested in its ramping PlayStation3, Xbox360 and server businesses, we still are not seeing the firm grow its gross or operating margins as previously expected, making us wonder if the firm’s 50% gross margin target is realistically achievable before royalty income ceases in early 2008.
• No change to calendar 2006 EPS estimate of $1.90, calendar 2007 EPS estimate of $2.15, and maintaining price target of $40 and HOLD rating. Our price target is based on a 20x multiple of c’2007 EPS on the core product business, plus the present value of royalties expiring at the end of 2007.
• Risks to attainment of our share price target include: Greater than expected chip pricing pressure, weaker than expected seasonal demand in 2H’06, less than expected inventory reductions in the distributor channel, greater than expected internal inventory builds, and weaker than expected gross margins.
• Consensus revenue estimates seem modest relative to historical seasonality, a positive: Consensus revenue growth expectations are fairly modest given the firm’s historical seasonality, and given its ramping gaming console designs, possibly setting up IR to beat expectations in the future.
• Valuation multiple sensitivity shows that IR could command a $50 stock price again if EPS estimates are too low; otherwise the stock is likely range bound near-term: We ran a scenario in which we increased revenues by four points versus our model and increased gross margin percentage by one-point versus our model, which drove about $0.50 of annual EPS upside for c’2007. Applying a 22x forward multiple to this new, higher EPS forecast would justify a $50 target price for IRF.
• IR’s Key MOSFET Patents Expire in 2008, a $0.40 Negative Annual Impact: These patents generate royalty and licensing revenues for IR of roughly $40 million per year, and represent a $0.40 negative annual EPS impact when royalties fully cease.
• June Earnings Analysis — Positives: 1) June revenues of +9% beat prior guidance of +5-8%. 2) September revenue guidance of +5-7% QoQ is ahead of consensus revenues. 3) Backlog grew +12% QoQ, and bookings grew +27% QoQ with ramping console and server platforms.
4) Distribution weeks of inventory fell slightly to around 12 weeks on hand. Negatives: 1) Gross margins are not expanding despite big revenue growth in June and expected in September. 2) OpEx guided to grow +5% QoQ, almost as much as revenue growth guidance. 3) Management again delayed its non-focus divestiture, not meeting its previous divestiture timing goals.
• In Summary: While we respect International Rectifier’s server and gaming console design wins, we remain at a HOLD on the stock given: 1) The firm is not making sufficient progress towards its 50% gross margin target. 2) IR seems to have made material margin concessions on its new gaming and server platform wins. 3) The firm is ramping up its OpEx spending significantly. 4) The stock’s valuation multiples are not compelling relative to peers or its own growth rate. Our $40 price target is based on a 20x multiple of c’2007 EPS on the core product business, plus the present value of royalties expiring at the end of 2007 (adjusted for interest income of $0.07 and net cash per share of $6.50). Historically International Rectifier has traded within a range of 13x to 31x its forward 12-month EPS estimates, with a median forward multiple of 19x. Our $40 price target is also based on a 2.2x enterprise-value-to-sales multiple (2006).

Investment Summary

Some Primary Investor Criticisms Regarding TRID are Overdone

• We were wrong to have been constructive on PXLW over the past year and are downgrading shares to HOLD given our view that the firm will not recover as much lost TV business in 2007 as we were previously hoping for. Design win traction with meaningful customers appears limited in its TV business, and we do not believe holding out hope that PXLW is acquired is reason enough to recommend the stock. We do still believe the firm has a compelling technology position and product roadmap, though we favor Trident Microsystems (TRID) in this space given their strong execution and earnings power. For those still in PXLW, there seems to be potential acquisition appeal from Broadcom (BRCM), NVIDIA (NVDA), TI (TXN), STMicro (STM), or LSI Logic (LSI).

• Microsemi delivered another quarter of solid execution and financial results, reinforcing our view that it is a compelling investment with significant growth opportunities, consolidation benefits, and factory efficiency benefits still left to reap; we also believe the firm’s unique characteristics provide chip investors with a relative safe-haven in times of a weak semiconductor tape, as well as capital appreciation opportunities during a stronger tape.

Several interesting comments were made recently that could drive a positive trading bias to the stock in the short-term including:

Highlights from the note include:

* There has been significant speculation that LSI Logic has won some low-end server business (Engenio) with Dell (DELL) and possibly with Sun Microsystems (SUNW) as well, a potential positive for the stock. Awaiting more visibility into the timing and magnitude of such design wins.

* SigmaTel reported Q1 results in-line with its earlier pre-announcement, with Q2 revenue guidance of $40-46 million in line with our prior estimate of $42.6 million. Execution issues continue with management citing product transition challenges, a soft end market, and market share losses to low-end competitors as primary factors.

* Zoran reported another quarter of significant revenue and EPS upside versus expectations, a positive that allows our growth and profitable diversification thesis further room to run.

* MSCC investor concerns regarding medical device market weakness are overdone, in our view: First, the FDA's recent decision to lower reimbursement rates for implantable defibrillators is unlikely to lessen demand from sick people that need the device. Second, recent weakness at medical chip customer St. Jude is more likely due to market share losses rather than just a weak market. Third, not yet reported Medtronic results and diminished transparency at Guidant (recently bought by Boston Scientific) are causing some investors to sell in the absence of greater visibility into that business. In total, we believe these issues have a worse bark than bite.

* Our EPS sensitivity analysis leads us to conclude that the worst of the medical device-related MSCC selloff is behind us with only a $0.03 potential negative EPS impact in downside scenario.

Samsung
Sigmatel, Inc. (SGTL)
Actions Semiconductor Co. (ACTS)
LSI Logic (LSI)

Berger and Schneider's favorite sector stocks -- in order of preference -- are : Genesis Microchip (GNSS), Monolithic Power (MPWR), ON Semiconductor (ONNN), Microsemi (MSCC), Zoran (ZRAN), Atmel (ATML), and LSI (LSI). Excerpt:

• Despite ATML’s run from around $2 in Q3’05 to current levels over $5.00, we think the company’s prospects for margin expansion and cost reduction, as well as healthy overall industry dynamics, will provide tailwinds to the stock over the coming months. Longer term, we see additional opportunities for cost reduction as the company has significant room to trim higher-cost manufacturing facilities and outsource the manufacturing of more of its commodity product lines.

* Q1 revenue pre-announcement again disappoints investors with management blaming weak sales of ATV chips in Europe and China; gross margins are also weaker than expected due to inventory reserves and poor new product yields.

* Investors should continue to BUY GNSS and can do so aggressively at or below $18 per share as we believe that the stock's fundamental thesis remains intact. Near term demand weakness and yield problems should be short lived, allowing a cheaper entry point into the stock, with good prospects for solid sequential revenue growth in both the June and September quarters, and likely at least $0.50 of earnings power in calendar 2H'06.