Friday, October 10, 2014

spending the money

There have been a few announcements over the past few months of big technology companies placing large sums of money betting on the future of computing, in particular the quantum computing variety. At first glance, it seemed to vindicate my entire career decision of getting involved in the physics found at the end of Moore's Law, with quantum computing my subdiscipline of physics for the past year (and nanotechnology being the previous one, so also yay!). But I find this raises some important questions, with answers to be eventually discovered over the long term, so the more important to start thinking about them sooner.

We'll begin with Microsoft, who made a splash into the scene of quantum computing with the establishment of Station Q in 2006, a sort of home to quantum computing theorists and a network of experimental labs on retainer. Recently they're in the news again for their continued investigation into the a <deep breath> topological quantum computer based on the ν = 5/2 fractional quantum Hall state </inhales>. This is the sort of thing that was heavily researched in the 70's and 80's and netted a couple Nobel prizes in the process, mostly owing to the advances in materials science that allowed growth of high-purity AlGaAs/GaAs films (you'll find transistors made out of this in your cell phone). Quick explanation: these films act as a two-dimensional conductors in which the collective behavior of electrons mimic particles that possess charges that are a fraction of the electron, which is neat because fundamental particles with this property don't exist. Those with the specific fraction of 5/2 (and at least another one, I'm pretty sure), have world lines that get all tangled up in knots, hence the 'topological'. And this actually improves the lifetime of the quantum information considerably, assuaging one of the major obstacles in building quantum computers. Theorists in particular love this implementation of quantum computing, and if you look back at the sum Microsoft is spending on this ($10 million, FOR REALZ?), you'll see that's all they can afford. No big fancy labs are being built with that kind of money, and Microsoft has no experimental effort to support experimental research. As a substitution, they do provide grants to experimental endeavors, with the Kouwenhoven group at Delft University in the Netherlands being the most notable. And actually they've been researching a related quasi-particle called the Majorana fermion, which has been a hot area of condensed matter physics research lately. What's not been so hot recently, however, is research into quantum Hall physics, with this guy apparently leading the experimental charge to understand the ν = 5/2 fractional state at Bell Labs (in Murray Hill, NJ), which hasn't been a hot place to work for 20 years. This is interesting because I know nobody working in this experimental subfield and, perhaps recently, considered every research group on the East Coast appropriate for the condensed matter experimentalist that I am (was?) without happening upon this group.

If dropping serious bones is your kind of thing, then IBM takes the cake with the announcement of three billion (!) invested in the future of computing. While supporting many efforts besides quantum computing, including nanophotonics, carbon in both graphene and nanotube form, and low power transistors such as tunnel field effect transistors and III-V nanowires, one can see IBM is serious about superconducting qubits by hiring former members of the Yale and UCSB quantum computing groups and putting them in PR videos. IBM certainly has the research facilities for experimental advancement in this fields, with a group at Watson Lab (Yorktown Heights, NY) leading the way, both experimentally and theoretically. Superconducting qubits have the distinct advantage of their fabrication being compatible with existing silicon semiconductor fabrication methods, and that you get to design the parameters of your qubits rather than dealing with whatever you can trap. But with Big Blue seemingly getting out of the hardware fabrication business, there are questions about its commitment to actually making things (as opposed to providing solutions!). But I suppose the superconducting qubits are easy enough to make, and there's always value in the talent and patents. Sweet delicious patents.

Google is known for making big bets on nascent technologies, and perhaps slightly less well known than the self-driving car, last year the Googs went in with NASA on a D-Wave machine. This purchase went over with much fanfare, including this past Feburary's cover of Time Magazine, but since I'm discussing quantum computing, I will stop talking about D-Wave right now. More far-reaching, however, is Google's hiring of John Martinis, the leader of one of the most prominent superconducting qubit groups in some sort of new arrangement I can't say I've seen before.

Quantum computing is one of many solutions we'll see in response to the expiration of Moore's Law in the next decade or so, when the number of transistors on a chip no longer doubles every 18 months. Concurrently there seems to be a change in the nature of scientific funding, with corporations providing grants to academics directly, previously the purview of government agencies and more recently public-private partnerships. Although this could just be news to me as a physicist; I seem to recall Intel sponsoring parts of the ECE department at my alma mater. The hybridization, if you will, of the Martinis group at UCSB with Google is definitely something to follow, especially since it's not so clear at this point what is going on. Seemingly contradictory to these splashy announcements is Microsoft's recent shuttering of their Silicon Valley research facility, suddenly and without warning, and the fact that IBM's committed 3 billion is actually a maintaining of current levels of funding for chip development. Which begs the next question: with these large companies doling out the cash on scientific research, what do they expect in return: PR, patents, talent, a working quantum computer...?

No comments:

Post a Comment