I wonder if YC would fund a new Bitcoin mining ASIC startup. I often regret not being able to pivot my previous company from FPGAs into ASICs :/
If you want to make one let me know, I'll happily be a tester for you!
YC has already funded at least one Bitcoin startup, which is currently doing quite well for itself (Coinbase).
Of all the ASICs that could be built, mining ASICs are the least risky. Their logic is simple, repetitive, and self testing. They have the highest yield and are the most flexible for binning.
And cryptocurrency mining has a _long_ history of pre-sales. That effectively eliminates all risk.
This is completely ridiculous. I've never seen anyone supply numbers to remotely back this up. A few people have linked to a blog post that through extrapolation and gross misunderstanding asserted that bitcoin mining used up as much electricity as all of Ireland (lets use our very best judgement).
The truth is the financial industry is 8% of GDP. How much power does it take to air condition all the banks of world?
Beyond that there is the fact that not all electricity has significant environmental impact and the fact that bitcoin mining ends up happening in places where electricity is the cheapest and most plentiful, which means it probably is not being generated with coal or oil.
This assertion is an enormous detachment from reality based on gut feeling.
The Bitcoin network hashrate is currently at ~3,800,000,000 GH/s. The most efficient miners are at ~0.1W/GH. So the Bitcoin network, at its most efficient, would be using 380,000,000 Watts right now.
I'm not sure what to compare that against. I looked up and found this article: https://www9.nationalgridus.com/non_html/shared_energyeff_of...
That article says that office buildings use ~1.53 Watts per square foot to cool the building (that's an average 1.53 throughout the year). So the power used to secure the Bitcoin network is equal to cooling ~248 million square feet of office space. Sounds like a lot, but pulling up a random office building, 55 Water Street, and I see it's 3.5 million square feet.
The U.S. as a whole uses 446,689,497,716 Watts on average. So Bitcoin is using 0.085% of the total U.S. power consumption. Or, 0.015% of world power consumption.
I have no idea if what we are currently spending on the Bitcoin network, in terms of natural resources, is more or less efficient than our current banking system. But those are the numbers.
Personally, I don't think any of its relevant. Reducing energy consumption should not be the primary focus of our species. We should focus on increasing sustainable energy _production_. Energy production is a core attribute of our economy and our civilization. It may even be the _most_ important thing to us, as a growing species. Obviously reducing waste when we can is good, but I'd rather expend the majority of our resources building out solar installations rather than quibbling about whether Bitcoin is power efficient or not.
They conclude that "the power currently used for Bitcoin mining is comparable to Ireland’s electricity consumption."
[0]: https://karlodwyer.github.io/publications/pdf/bitcoin_KJOD_2...
Before I point out the absurd flaws in how they came to this conclusion, how likely do you think it is that in 2014, the electricity use for bitcoin mining was equivalent to all the private, public, commercial and industrial electricity used by over 5 million people?
Undoubtedly there is huge amounts of energy being pumped into Bitcoin daily, but it's environmental effects seem over-hyped compared to many other human endeavours.
If that's accurate, then that's not the same as building a mining chip. The mining, for them, was just a way to make their microservices stuff viable. Everyone I ever heard who discussed those offerings of theirs knew that the tiny mining cores weren't profitable, and said as much. It's obvious that any chip which isn't 100% dedicated to mining won't be economically viable.
https://www.blocktrail.com/BTC/pool/21inc
The chips on devices thing was a pivot once their initial idea didn't work.
What level of capital commitment would you need to build a proof of concept?
Yes, I co-founded a company a few years back that designed, built, and sold FPGA mining equipment. We are probably one of the very few hardware Bitcoin mining companies to ship hardware on-time with nothing but happy customers. I designed the FPGA firmware (which is open source), embedded firmware, and drivers. Others designed the boards, software, etc.
> What level of capital commitment would you need to build a proof of concept?
Depends on what kind of proof of concept. A small scale proof of concept, where you have a board in your hands that's mining Bitcoin, would cost ... nothing. I'd just lend my devkits to the company :P Or the existing FPGA miners that we built.
Sometimes it's required to simulate a significant percentage of the final ASIC design. That would require a much higher end FPGA devkit; ~$10k-$20k. Sometimes you can get FPGA vendors to loan you the kits, though.
If the goal of the company is merely to sell mining equipment, or if that's necessary to promote a crowd funding round, then the proof of concept should probably be closer to what that shippable product would be. I'd rough that out to maybe $100k to pay a board designer to do layout and fund a spin or two of small quantity prototypes based on FPGAs for simulation. That board design would then be re-used once the ASICs are ready. That's back of the napkin; could be more or less depending on scope but generally in that 6 figure ballpark.