Wow! This is a week to remember — the biggest breakthrough in the production of semiconductors in 40 years — real Nobel Prize stuff. I moved to California from the East Coast in 1968, a year before a private company named Intel began using polysilicon to make transistor gates, which determine whether the transistor is on (1) or off (0). In 1970, our investment management company made a venture investment in Intel, and I have followed the company ever since.

Polysilicon is the material that Moore’s Law is based on — Gordon Moore’s prediction in the mid-’60s that chips would shrink 50% every 18 months. The roadmap to continue that shrink rate until about 2010 has been known for years, but most people thought it would be very difficult to get past that point using polysilicon.

A couple of years ago, Intel told me that they thought they had a technology to extend Moore’s Law for at least another decade, and this week they revealed it. In fact, Gordon Moore, now 78, flew in from his Hawaii retirement home to label this the “biggest change in transistor technology” since polysilicon.

What did they do? They replaced polysilicon with a metal gate based on hafnium, which holds a charge better than polysilicon and therefore can be shrunk smaller before the gate leaks electricity. Transistors leak more current as they get smaller, which means they also overheat. The new technology leaks only 10% to 20% as much as the old, so your batteries will last much longer. Then they combined the metal gate with a higher insulating material (called high-k) to create a more efficient structure. The chips run faster and, importantly, use substantially less energy.

IBM has had a team working on the same idea, and both teams announced results last Friday — in good geek style — too late in the day to have much news impact. The difference is that IBM builds gates inside the wafer, while Intel builds them on top, which makes it easier to produce. Consequently, Intel’s technology will be on the market by the end of this year, showing up in servers, desktops and laptops, while I don’t expect to see commercial products from IBM until 2008 or 2009. So, when IBM will just be starting to put their chips on the market, Intel will already be in full production in three factories in the first half of next year. For a rare 40-minute video visit inside one of Intel’s new fabs, click here.

Intel’s new high-k metal gate chips, code-named Penryn, will have 410 million transistors at the next technology point of 45 nanometer line widths. The current polysilicon chips have 280 million transistors at the same line width. Over 2,000 transistors that are 45 nanometers wide can fit on the period at the end of this sentence.

So, Michael, if it is such a great breakthrough, when do we buy back INTC stock? I think the best answer is “soon.” The Windows Vista introduction on Tuesday was not an “event” that will cause a sudden jump in seasonally-slow PC sales, so the March quarter will probably be as lackluster as Intel has already indicated. But everyone already knows that, and I think what is not priced into the stock is the substantial market share they can win back from Advanced Micro Devices in the second half of the year, based on the Penryn processors. These will be high-end, high-profit, server-class chips. At this point, it looks like we have avoided a sharp market drop (see the Market Outlook below), so if the rally that started today is a real breakout to the upside (confirmed by a resumption of the dollar’s decline), I will pull the “buy” trigger. We may buy LEAP calls when the time comes — I will send you a Flash Alert when you need to take action.

More of our companies are reporting December-quarter earnings or other news, and it looks like the MegaShifts that I have identified will be the market leaders in 2007. Here are the details.

Avian Flu MegaShift

Gilead (GILD) reported another excellent quarter. December fourth-quarter sales hit $899.2 million, up 47.6% from the December 2005 period and well ahead of the $847.7 million consensus. They did 78 cents a share, a dime better than expected. The new Truvada HIV drug accounted for $327.1 million in sales, $31 million above consensus expectations. On the conference call, management said that the company will no longer break out its forecast for HIV drug sales, but gave an overall 2007 revenue forecast between $3.4 billion and $3.5 billion, up 31% to 35%.

The stock shot up over 11% today to an all-time record high of $71.53, and out January 2008 $50 LEAP call options (YGDAJ) climbed 35% to $24.95. That’s better than a sharp stick in the eye! Continue to hold YGDAJ for my $30 target.

Biotech MegaShift

Affymetrix (AFFX) reports next Wednesday after the close. They already preannounced December-quarter revenues around $100 million, up from $84.7 million in the third quarter and just above the consensus for $99.5 million. AFFX showed growth even though they didn’t launch their new GeneChip, which has 500,000 arrays on a single chip, until December. There were costs associated with the launch, so it is possible that earnings will be a penny or two under the consensus estimate of four cents a share. The range is from a loss of a penny a share to profits of 10 cents a share, so there really is little consensus on what happened with the GeneChip launch.

Guidance will be more important, because expectations are low for the March quarter — $89 million in sales and one cent a share, with a range from minus a penny to plus two cents. But the launch of the SNP 5.0 Array chip could create revenue and earnings upside for the guidance, with improving profit margins every quarter in 2007.

I was surprised to learn, this week, that one of the new short sale newsletters is targeting AFFX, just as the company is beginning an important new product cycle. I don’t know what the newsletter is thinking, but I think they are about to get squeezed. Buy AFFX before the earnings report if it dips under $24 in a market-related decline. My $40 target is still good, and may even be low.

Millennium Pharmaceuticals (MLNM) won’t report for a couple of weeks, but they did say that they grew product sales 15% to $220 million and hit their 2006 goal of non-GAAP profitability for the year. That was pretty good considering Celgene launched Revlimid in mid-year for the multiple myeloma indication that Millennium’s Velcade targets. Millennium is spending a whopping $400 million a year on R&D to expand the label for Velcade, and, of course, the current revenues do not reflect any results from that spending yet.

The company guided for 2007 Velcade sales of $240 million to $260 million, indicating growth of only 9% to 18%. But the first half again will be comparing against a period when Revlimid was not taking significant market share, so that is not bad. On even that modest growth forecast, plus royalty income on Integrilin of $140 million to $150 million, the company expects non-GAAP profitability in the $10 million to $20 million area. With more than $800 million in cash, they have the resources to expand Velcade indications dramatically. MLNM remains a buy under $11 — it is trading back and forth across that level — for my $23 target, possibly via a takeover.

China MegaShift

UTStarcom (UTSI) won more IPTV (Internet Protocol TV) contracts with China Telecom and China Netcom, the two largest network providers in China — actually, two of the largest in the world. UTSI is a major supplier of IPTV equipment in China, and over the next few months that will become the “story” Wall Street needs to return to the stock.

I got an email from Jack about UTSI: “UTSI is one of the biggest customers of Sigma Designs (SIGM) for video decoders, where they have a monopoly position for HDTV decoders. Why don’t you recommend SIGM as another way to play both China and Content on Demand?”

Sigma Designs has a leading position in video decoders, but these chips decode well-known standards like MPEG-2 and MPEG-4. Other companies like ST Micro and Broadcom already ship video decoders, and I expect Cisco to start making their own chips this year. Cisco acquired Scientific-Atlanta, an important Sigma Designs customer, which has a long history of buying chips from outside suppliers at first, and then developing their own chips to replace the outsiders as volume picks up.

Like flash memory, the demand forecast for video decoders looks great. But demand attracts supply, and just as SanDisk recently tanked in the flash memory area, Sigma Designs probably has one or two good quarters ahead before substantial new competition takes chip prices down towards commodity levels. I don’t see how we could hold this stock for the full year required to get capital gains tax treatment. So, I don’t see it as a long-term value taker in the industry, like UTSI or Harmonic. Buy UTSI on dips under $9 for my $15 target.

Content on Demand MegaShift

Harmonic (HLIT) reported an excellent quarter and won an Emmy! The quarter was well above the consensus at $75.3 million in sales and 13 cents a share. If you look at revenue for the last three quarters, the pattern is pretty nice: June, $53.3 million; September $62.9 million; December $75.3 million.

December-quarter revenues were strong due to cable orders, and satellite orders were strong but came in at the end of the quarter. The company’s book-to-bill (orders to shipments) ratio hit 1.20, or 20% more orders than shipments. Many of those late-in-the-quarter orders will ship in the March quarter. Meanwhile, the cable companies were focused on getting VOIP (Voice over Internet Protocol) telephone services going for most of 2006, but then realized Verizon was starting to eat their lunch in video and came back to ordering from Harmonic in the December quarter. This sets up an interesting opportunity for the March quarter to show strong satellite sales of very profitable MPEG-4 video encoders while cable orders continue strong. I think it will happen, and Harmonic will beat analyst expectations and their own cautious guidance again, guide above expectations for the June quarter, and say the second half of 2007 will show good growth from the first half. Sales to satellite customers are expected to triple this year from 2006.

If all that happens, HLIT will do $300 million and earn 45 cents a share in 2007, far above the $273 million and 30-cent consensus estimate, and the stock will be in the mid-’teens by the end of the year. I am raising the HLIT buy limit to $9 and the target price to $16 for this year. When Wall Street realizes the video build out will last until 2010, HLIT should go much higher.

Now about that Emmy. Harmonic and Time Warner, a major customer, shared the Science, Technology and Engineering Emmy for Time Warner’s Start Over video-on-demand service. Harmonic’s DiviCom Ion encoders are used by Time Warner to efficiently deliver high-quality video streaming for the service.

Comcast (CMCSA) dropped sharply today in a great example of how short-sighted Wall Street is and how it is making it harder for U.S. corporations to manage themselves for the long term. The company reported revenues up 30%, including some from the Adelphia buyout, to $7.03 billion. That was about equal to the consensus for $7.05 billion. Earnings hit 20 cents a share, short of the consensus for 24 cents, but the difference was almost entirely from accounting adjustments related to updated valuations of the Adelphia and Time-Warner transactions. That is not what hit the stock today.

In the December quarter, Comcast booked 613,000 digital video subscribers, 508,000 digital phone customers and 488,000 cable Internet subscribers. Remarkably, they added 110,000 new customers in basic video — their strongest growth in 10 years — and this was almost certainly related to people buying the “triple play” offering of basic cable, phone and Internet access. As those folks upgrade their TVs, they can be moved to digital video subscribers. For all of 2006, Comcast added 80,000 subscribers net to basic video, and they expect to beat that in 2007.

For this year, Comcast is looking for cable revenue growth of 12% and overall revenue growth of 11%, both in line with the consensus estimates. None of this is what hit the stock today, either.

Wall Street’s “problem” was that Comcast said that they will spend $5.7 billion on capital spending, well above the $4.8 billion consensus forecast. Consequently, free cash flow will be flat in 2007, after 30% growth in 2006. They will use the extra $900 million in spending to accelerate the introduction of new high-definition video products and to accelerate selling services to small- and medium-sized companies. That’s what hit the stock today.

But why? If Comcast had said that they were increasing their stock buyback program by $900 million, the stock would have gone up today, not down. Instead, they are investing in HDTV and the large, previously overlooked business market. Is this a bad idea? I don’t think so — both of these areas will produce higher returns on investment than just buying back the stock.

In the long run, this won’t matter, but in the short run, it really bothers me that Wall Street would ding a company for investing in its future instead of buying back stock. Management essentially thumbed their nose at Wall Street and said that they will split the stock 3-for-2 on February 21. Well done. CMCSA remains a good hold for my $62 target.

Market Outlook

The S&P 500 broke out over my 1440 target today in decisive fashion, and it is time to get fully invested. The next stop is in the mid-1500s, and from there this upturn could expand into something really big. The underlying driver has nothing to do with housing, consumer debt, oil prices, or anything else you probably are worried about. The driver is simply that the Fed is aiming a fire hose of liquidity at the economy, and it usually goes into financial assets first, before spreading rapidly through the rest of the economy. The 3.5% preliminary GDP report for the December quarter is testimony to their success.

Will inflation pick up? Of course. Even with the constant redefining of the Consumer Price Index, they won’t be able to hide it. So the Fed will raise interest rates, not cut them, but they will continue to print money. Destroying the value of the dollar is the only way to get out from under foreign holdings of U.S. debt, overleveraged real estate and incredible Social Security and Medicare obligations. The plan is that the debt holders and retirees will get paid, but they’ll get paid in rapidly depreciating dollars, so the U.S. economy won’t be hurt. The plan will even work, for a while. If the dollar is falling, real assets — stocks, oil, real estate, gold and other metals, and so on — must rise.

So, what should you buy? It is always a good idea to diversify more, and even though it is very tempting to average down in some of our stocks, especially in the New Energy Technology MegaShift, I recommend buying a different stock in the same area you want to build up. Right now, I think you should especially target these stocks:

• Avian Flu – Top Buy BioCryst (BCRX) first, but if you are already loaded on it, Crucell (CRXL) is OK for new purchases.
• Biotech — Top Buys Dendreon (DNDN) and eResearch (ERES) first, then the Biogen-Idec January 2008 $45 LEAP calls (YZUAI), Geron (GERN), and Millennium under $11.
• China — UTStarcom.
• Content on Demand — Certainly Harmonic under my raised buy limit, Silicon Image (SIMG), and Top Buys Telkonet (TKO) and Zhone (ZHNE).
• New Energy Technology — Start with the petroleum-related recommendations: Top Buys Connacher Oil & Gas (T.CLL), Gasco Energy (GSX), Infinity Energy (IFNY) and Rentech (RTK). Then add the fuel cell companies, Fuel Cell Energy (FCEL) and Plug Power (PLUG).
• Robotics – iRobot (IRBT).
• Security — American Science (ASEI).
• WiMAX — Top Buy Alvarion (ALVR), then Airspan (AIRN), Terabeam (TRBM) and MobilePro (MOBL).

I expect to send you a Flash Alert early next week on one or two more immediate buys.