The IMF Outlines Its Peculiar Assessment of Cryptocurrencies
Follow Yassine on Twitter @yassineARK
Christine Lagarde, Chair of the International Monetary Fund (IMF), detailed the case for central bank digital currencies (CBDC) at a Singapore Fintech conference earlier this week.
In her speech, Lagarde evaluated the role that central banks will play in the evolving financial landscape. One option will be a central bank digital currency (CBDC), "a state-backed token, or perhaps an account held directly at the central bank, available to people and firms for retail payments.” At the highest level, a CBDC would be digital money issued by a central bank to serve as legal tender, similar to cash, primarily for retail payments.
Lagarde delivered her speech two days after the IMF released a research paper highlighting many of the points she made. Providing some context and perspective from that paper is the table on page 39, delineating the IMF's view on various forms of money compared to potential CBDCs.
Most interesting to us is the IMF’s inaccurate representation or misunderstanding of cryptocurrencies. In particular, its assessments of "scalability", "theft and loss risk", "transaction costs", and "anonymity costs" seem misplaced. On scalability, Bitcoin's base layer is a settlement layer for large value transactions, as opposed to the IMF’s suggestion that it is unsuitable for large payments. On anonymity, Bitcoin transactions are pseudonymous and transparent, not anonymous. Finally, on transaction costs, while the IMF suggests that the demand for energy will scale with the number of transactions, the fact is that the energy necessary to secure the network is independent of the number of transactions the network can support.
While the report conveys that central banks are grappling with changes in the financial landscape, ARK believes that the IMF must be more open-minded about the contenders for “money” before proposing a solution.
Nvidia Expands from Deep Learning to Mass Market Machine Learning with RAPIDS
Follow James on Twitter @jwangARK
While the financial community fixated on excess inventory of graphic processing units (GPUs) after Nvidia reported weak third quarter sales, we believe that its announcement of RAPIDS—a GPU framework for machine learning—was far more interesting and will be much more impactful long term.
Nvidia pioneered the concept of using GPUs to accelerate deep learning applications, now a $3+ billion business. Deep learning, however, makes up only a fraction of the machine learning market, which includes dozens of different algorithms and software frameworks. While leading edge internet firms use deep learning and neural networks, for example, most large enterprises use some variant of Hadoop for data analytics.
Nvidia’s RAPIDS framework brings the power of GPU acceleration to common machine learning and data analytics frameworks today. Unlike its foray into deep learning, where it created the market from scratch, Nvidia launched RAPIDS which taps into the large and growing market for analytics. With Moore’s Law grinding to a halt, we believe Nvidia’s move could not be more timely.