My first reaction is that it's way too optimistic: full quantum computers are really hard, slow developing, and not actually that useful since aside from famous applications like integer factoring (which as far as I know, is purely destructive - of security) there aren't that many speedups available from algorithms: https://en.wikipedia.org/wiki/Category:Quantum_algorithms Factoring 15 isn't an impressive achievement.

Then I notice the caveat 'for discrete optimization', and that it's D-Wave stuff. I'm not actually sure what D-Wave is doing or what it'd be used for. I look at http://en.wikipedia.org/wiki/Quantum_annealing and it says it's an optimization process like simulated annealing (OK, I know what that is) which uses quantum tunneling for the annealing jumps (not sure why that's better than regular RNG-guided jumps across fitness landscapes but I'll roll with it). What's the speedup here? "This O(N^{1/2}) advantage in quantum search (compared to the classical effort growing linearly with \Delta or N , the problem size) is well established".

So it's a linear classical algorithm versus a sub-linear square-root (faster growing than O(log N) if that helps). This is... OK but it's certainly no Shor's algorithm, turning an exponential into a quasi-log complexity operation! And then there's the question about how much the extra accuracy is actually worth. (If the difference between the answer returned by your best classical computer doing N ops and the quantum chip which can do N^2 equivalent ops is 0.00001%, this may or may not matter in practice.) This summary isn't consistent with the 'Scaling Benchmark - QC vs Intel', which is claiming advantages which increase faster than a square (300 qubits = 500x; 400 qubits = 1,000,000x; 512 qubits = 10,000,000,000; but 300^2=90,000; 400^2=160,000; 512^2=262,144) so I'm obviously missing something or doing something wrong. Between these 3 large issues, I can't be confident of any predictions about the value of future D-Wave chips. Well, let's take that at face-value.

Business-wise, they seem to be in a little trouble, dependent on massive external investment (eg. http://www.technologyreview.com/news/429429/the-cia-and-jeff-bezos-bet-on-quantum-computing/ ) with their current product costing $10m and the one published example of use being questionable http://blogs.nature.com/news/2012/08/d-wave-quantum-computer-solves-protein-folding-problem.html :

> The model consisted of mathematical representations of amino acids in a lattice, connected by different interaction strengths. The D-Wave computer found the lowest configurations of amino acids and interactions, which corresponds to the most economical folding of the proteins. It worked, but not particularly well. According to the researchers, 10,000 measurements using an 81-qubit version of the experiment gave the correct answer just 13 times. This was owing, in part, to the limitations of the machine itself, and in part to thermal noise that disrupted the computation. It’s also worth pointing that conventional computers could already solve these particular protein folding problems.

This suggests to me that they're already having serious scaling issues. Or http://www.technologyreview.com/news/424163/tapping-quantum-effects-for-software-that-learns/2/

> "You send in your problem and then get back a much more accurate result than you would on a conventional computer," says Rose. He says tests have shown software using the D-Wave system can learn things like how to recognize particular objects in photos up to 9 percent more accurately than a conventional alternative. Rose predicts that the gap will rapidly widen as programmers learn to optimize their code for the way D-Wave's technology behaves.

I'm actually reminded a lot of Symbolics and Thinking Machines, reading about D-Wave.

So, existing predictions:

1. Intel selling QC by 2025: http://predictionbook.com/predictions/4444 4%
2. >100 factored by 2020: http://predictionbook.com/predictions/3213 40% (may already have happened)

    By 2030: http://predictionbook.com/predictions/3211 55%
    By 2045: http://predictionbook.com/predictions/3212 75%
3. >100 qubit system by 2011: http://predictionbook.com/predictions/1502 1%, but I judged this as happening because D-Wave sold the Lockheed system in the nick of time. Embarrassing. I should have specified a fully-general QC chip.

So, new predictions. My big concerns are business viability and scaling:

1. D-Wave will be bankrupt, bought, or defunct in 5 years: http://predictionbook.com/predictions/8566 30%

    * in 10 years: http://predictionbook.com/predictions/8567 50%
2. D-Wave will be commercially selling a >=1024-qubit quantum annealing chip by 2015: http://predictionbook.com/predictions/8568 40%

     * by 2020: http://predictionbook.com/predictions/8569 60%
3. D-Wave will be commercially selling a >=2048-qubit quantum annealing chip by 2016: http://predictionbook.com/predictions/8570 25%

    * by 2018: http://predictionbook.com/predictions/8571 35%