How is the hotel/touristic industry?

How would the semiconductor and software development industries compare to each other 10 to 15 years down the line in terms of pay, industry size and consolidation (job security)?

• The semiconductor industry has defined most of the breakthroughs in the past 50 years. Yet the cyclical nature of the industry coupled with a capital intensive nature means there is very less chance for new entrants and a lot of consolidation among the companies that are able to survive. This also means pay packages and hiring in the semiconductor industry has not kept pace with software development and related roles for some time now. Hardware companies are known to go bust when banking on the wrong products and end up firing a lot of talented and experienced engineers. On the other the current demand of software engineers is at an all time high. They are paid like never before and the number of opportunities too are unmatchable. Also the skills needed for software companies while intellectually demanding, still can be relatively easily picked up at the entry level (especially compared to jobs needing intensive engineering skills like that of electronics and chemical engineers, etc.) This has also led to a lot of talented  people from non-computer science backgrounds to move to this field. 1. How long is this bull run likely to last for the software companies before they go through similar consolidation phases? Is it going to be 10 years from now, or will it only be after 30 years from now? 2. With Moore's Law tapering off, how will semiconductor companies cope in terms of innovation? Will chip design jobs be even needed a decade from now (on a large scale)? 3. Most importantly, for someone with 2 years of industry experience in semiconductors along with it being my passion since at least a decade (somewhat satiated now :-P), does it make sense in staying put here or moving to software development?

• Answer:

  Semiconductors companies will continue to consolidate, especially people doing pure digital or mostly digital design. There's still a little room for folks making sensors on so on to start small companies, but digital is basically done - you can buy a chip with 4 CPUs, a graphics processor, audio and a whole bunch of other gubbins for a few bucks, and that's far more than you need for most applications. We're at a slightly ridiculous point now, where one companies "custom" ARM based SoC can run code written for another, because they're built out of all the same building blocks, right down to the address map. As a supplier to semiconductor companies, we're down to 4 make-or-break customers now, and maybe another 6 or 7  that are significant. In parallel to further consolidation, the gains to reducing feature size are going to evaporate in the next 3-5 years, and designers will look to better design techniques, better EDA software, better automation, and cost reductions for benefits. That's going to be exciting - there's a lot of interesting ideas out there that have been roadkill on the path to the next technology node - but unless there are astounding cost breakthroughs, its not going to stop the consolidation because semi-conductor design and manufacturing and capital intensive businesses. I actually think this will be very good for EDA and for designers - some creativity will re-enter the business and there'll be players big enough to pay for it. Software has very different economics. Its very cheap to make, but generates very unpredictable returns on that investment. So I don't think consolidation is very likely. Google, Amazon and Facebook have some advantages because they know how to operate extremely large server farms and write software for them, but they've also basically comoditized access to these facilities. Its not like Intel's design and process methodology where no-one, not even most Intel employees, knows the whole picture. If you pay a very small amount, you can deploy a website to millions of users as if you owned the infrastructure yourself. Software works something like Holywood - risk-free stuff is done by big companies, startups take the risks and generate the returns. If you have 2 years experience in semiconductors, you're basically still entry level and would be in software too. At entry level, software companies currently pay more because they need raw brain-and-hand-power desparately. At more senior levels, semiconductors still pays more and I think it always will - the risks are so much bigger, the investment in employees has to be comparably bigger.