Year to date, the best-performing sector of the S&P 500 is technology. Can tech continue to run wild?
The tech sector is up 22% year to date. By contrast, telecom is down 9%.
I have been bullish on many of the top names. For example, back in March, I told investors not to forget Micron because it was going higher. I thought Micron's move away from commodity memory toward more complex types of memory would bode well for the company. Strong demand with big increases in gross margins has been driving Micron higher. Cloud computing is eating up memory. Cloud computing is expected to be a $5.2 billion opportunity for the memory makers. Micron's 3D NAND chips offer 50% more capacity than the competition and enable customers to store 8 terabytes (TB) in just 2.5 inches of space.
In April, I thought Lam Research could continue to take market share. Lam is a huge beneficiary of the growing memory market, especially the transition to 3D NAND. Furthermore, chip makers are rushing to convert their fabs to 1x DRAM (i.e., 10 nanometer DRAM), since 20 nanometers, first produced in 2014, are getting long in the tooth.
Lam believes that from 2016 to 2020 the semiconductor industry will spend a staggering $50 billion on wafer fab equipment to manufacture 3D NAND memory. Lam thinks it can take half the growth in the market. The worldwide semiconductor market is expected to grow between 5% and 7% in 2017 to $358 billion, while semi equipment capital expenditure is likely to grow about 4%.
In August, I turned cautious on shares of fourth-best performer Nvidia Corp. (NVDA) . I thought its valuation was stretched after the run the stock had. The stock gave back a little, but it still is up 68% year to date. I still like Nvidia and think it will get its legs back. (Nvidia is part of Jim Cramer's Action Alerts PLUS charitable trust.)
Finally, one of the most overlooked parts of the software business is something called electronic design automation, or EDA. Cadence Design Systems Inc. (CDNS) and Synopsys Inc. (SNPS) have been hitting it out of the park. Cadence is up 57% and is the seventh-best performer in tech land, while Synopsis is up 38% and is the 18th-best-performing stock. As the semiconductor industry moves to smaller dimensions, the industry has come to rely on the EDA industry for verification tools, its gigantic storehouse of intellectual property and its design software. Without software to design semiconductors, chips wouldn't exist. After all, it's impossible for Nvidia to design, verify and simulate a Pascal processor with 15 billion transistors squeezed on to a 600-millimeter square die without software.
Because of a strong upgrade cycle, I thought Synopsis would reach $70 a share. I was wrong. The stock closed at $81 yesterday.
I think the trends that propelled these names can continue. For example, at the Flash Memory Summit, Lam Research hiked its five-year NAND wafer-equipment spending estimate from $50 billion to $70 billion over the next five years.
At the end of September at its analyst day, Applied Materials Inc. (AMAT) said semiconductors make up more than 25% of all electronic components. From 2013 to 2017, Applied Materials estimates the equipment market will grow from $36 billion to more than $60 billion. The average gigabyte per smartphone is expected to jump 140% from 2015 to 2018. The company believes semiconductor makers are thinking about 7-nanometer nodes, which could be twice as expensive to manufacture as a 28-nanometer die.
Nvidia's Tesla V100 chip is expected to have 21 billion transistors and more than 100 billion vias (vias are interconnections between different layers of a chip). That is going to be one really expensive chip to make.
The OLED (organic light-emitting diodes) display industry is expected to invest $18 billion per year from 2017 to 2020 because of advancements in display technology. That means the equipment market will grow an estimated 425% as display makers switch from manufacturing LCD (liquid crystal display) to OLED displays.
The memory and logic to drive artificial intelligence (AI) is staggering. An AI server requires four times more memory and eight times more logic by area than the current enterprise server set-ups.
While valuation is always a concern, for long-term investors I would look for sell-offs and earnings disappoints to add to positions.