With the rapid advancement of Artificial Intelligence (AI), we have seen an exponential increase in the number of brands that look to provide some product or service related to AI.
Just looking through Battalion’s AI watchlist, the statement above seems accurate. We have consultancy brands like IBM (IBM) and Accenture (ACN); AI and analytics platform providers like Cloudera (CLDR) and Salesforce (CRM); Veritone (VERI) who has deviated from the pack developing an AI Operating System and finally, we have the AI chipmakers such as Nvidia (NVDA), Xilinx (XLNX) and Intel (INTC).
What’s interesting though is that the discussion (on Wall St) around chipmakers is focused on processing power and their AI capability as we know it. The well-established AI use-cases such as autonomous driving, identifying cat pictures, intelligent robotics and figuring out which type of flower belongs to which species are at the forefront of analysts’ minds.
What I believe Wall Street analysts forget, is that once there are advancements in chip technology there will be parallel advancements in AI capability. Moore’s Law suggests that the number of transistors on a CPU would double exponentially every two years. Transistors are a semiconductor which sits on a computer chip through an integrated circuit and can effectively turn on and off. This leads to a computers ability to make calculations through binary calculations (0’s and 1’s).
Moore was probably right – think about the processing power that we had in the 1970s and what we had in 2014 and what we have now…(see the graph below).
Some analysts and technologists, however, have suggested that Moore’s Law is theoretically broken as we have reached a point where transistors cannot get smaller (think atomic particles here). Whilst they’re right – they’ve failed to consider the impact that alternative types of processing chips will be developed, enter Quantum computing and Neuromorphic chips.
Quantum computing uses Quantum Bits (qubits) as opposed to transistors. Qubits can be 0, 1, both or quantum superpositions – which sit between 0 and 1. With this flexibility, computations can be calculated quicker than ever before. Remember transistor-based chips can only be 0 or 1.
The other type of chip in development is called a Neuromorphic chip. The Neuromorphic chip is designed to mimic the human brain where neurons are sending signals to organs and muscle tissue. Neuromorphic chips have cores which have neurons and synapses and communicate through a way called Spiking Neural Networks. Effectively sending information between different Neurons. This process of information exchange is called a spike. So what’s the benefit? Very low-cost and very quick processing as energy is not being consumed when there is no spike.
Now how does this all tie into Moore’s Law? Well, let’s look into average power density per core multiplied by the average amount of cores per chip (alongside transistors). Again the trend line shows an impressive exponential increase in computing power.
So whilst Moore may be wrong from a certain point of view (yes transistors can’t keep exponentially increasing every two years), he was right though that the processing power is still continually increasing – it’s just reaching the end of one technology cycle and starting a new one.
“I predict we could see a continued exponential increase in computing power where chips become smaller, faster and much more energy-efficient”.
What happens when we move away from standard chips with transistors to technology such as Quantum or Neuromorphic? I predict we could see a continued exponential increase in computing power where chips become smaller, faster and much more energy-efficient.
As an investor, I’m looking to invest in businesses that are pioneers in the Quantum and Neuromorphic computing fields. From my research, three brands making strides are Intel (INTC), IBM (IBM) and Google (GOOGL). These brands are exciting innovators (yes, I know, I just said IBM), have great technical value indicators and have other revenue streams (should either of these technologies fail).
Remember to do your own research and consult a registered financial advisor before you make any investments.