4 ms·
As a EE myself, who left the hardware industry to become a software engineer, I agree completely. It's also worth examining the reasons why chipmakers in today'
by whack 9y ago
As a EE myself, who left the hardware industry to become a software engineer, I agree completely. It's also worth examining the reasons why chipmakers in today's world can't ever compete with SW companies for top talent:
1. The market size and opportunity in SW is tremendous as compared to HW. SW is eating the world in nearly every industry, and this is going to continue for the next 10-20 years. HW used to be great, during the time of Moore's law. But with Moore's law coming to an end, it's going to be harder and harder for chipmakers to compete with the same products they released 3 years ago.
2. Entrepreneurial opportunities are severely limited in HW. A software startup can be launched by 2 college kids in a dorm room, using a couple thousand dollars of angel investments, and scaled nationally with 1-2 million in VC funding. With HW startups, because of the engineering/logistical nightmares involved in fabricating a chip and selling a physical product, you need so much more money/credentials/connections to get a startup off the ground. SW devs can realistically dream of building/joining a world-changing startup, but HW engineers are mostly resigned to being a cog in a big machine.
3. Paranoia and risk aversion. Strongly linked to the above, is the culture of rational risk-aversion that exists in any HW project. With a software project, the Agile/MVP methodology works extremely well, because the cost of making changes, or fixing prior mistakes, is so low. In HW, the costs of making changes and releasing a buggy product, are astronomically higher. Because of this, HW projects are extremely risk-averse. Junior developers are micromanaged by their superiors. People are given little autonomy to innovate and try new ideas. Project timelines are measured in years, not months. From an artistic perspective, developers love having the freedom to build freely and get their product in the hands of users as soon as possible. This is so much easier in SW as compared to HW.
Combine all 3 of the above, and it's easy to see why SW is so much more lucrative and enticing. As an EE myself, I tell everyone I know that unless they have a deep passion for the field itself, they should strongly consider a career switch.
- marktangotango 9y agoThe bug consideration is huge in hardware. One wonders what a different world it would be if the zilog z800, the most advanced 16 bit chip of its time, would’ve had fewer bugs.
- twtw 9y agoRe: Dead Moore's law and entrepreneurial opportunities. It seems to me that the fact that Moore's law has ended may end up with a flip side that could potentially be awesome. Previously, it nearly always made more sense to do something on a CPU and wait for performance to double than to develop a custom ASIC. With the free lunch deal over, I suspect that we will begin to see a ton more custom ASICs and architectural innovation. This has already begun for ML/DL/AI. Without Moore's law, the semiconductor industry may get a lot more exciting. It might also just fade to commodity even more, but that isn't as fun to think about.
- madengr 9y agoMoore’s law is not necessarily dead with respect to analog integration, and III-V semiconductors. Much room for growth. Xilinx has some amazing RFSoC; FPGA fabric with integrated GHz DAC & ADC.
- sigstoat 9y agoi'm long on semiconductors for this reason. there's a lot more that can be done with them than grinding out increasing dense general purpose CPUs and DDR memory. we saw a tiny bit of it with the introduction of MEMS IMUs, and we'll see more yet. sooner or later we'll get low costs on small quantity ASICs, and things will go really nuts.
- deepnotderp 9y agoDing ding. I've always been of the opinion that the end of Moore's Law will result in a new wave of startup innovation. When the cost of newer nodes are a barrier to entry and the advantages of startups can be obviated by just moving to the next node (which the startup cannot do), then innovation stagnates. This is largely what killed many FPGA startups (e.g. Tabula, which btw had super cool tech).
- madengr 9y agoI’m also an EE, but designing RF/Microwave hardware. Software, specifically SDR, is fun to do as a hobby, but I’d never want to do it for a living. Too much abstraction in software; libraries upon libraries. With thing such as antennas, and RF circuitry with a few transistors, there is almost no abstraction.
- twtw 9y agoWould you mind sharing what kind of applications you are working on? Defense, aerospace, mobile phones, other? I've heard worse things about the defense and aerospace industries than I have about semiconductor industry, so I'm just curious.
- madengr 9y agoAntennas, power amplifiers, up and down converters, etc, typical RF stuff, VLF through mmWave. Defense contractor in flyover country. The pay is good as there are few who do this stuff. Almost 22 years at the same company. Totally opposite of the startup hopping software whiz.
- ultrasounder 9y agotouché! Also on the same boat pivoting out of the H/W Product development hell