In 1951, Martin Minsky and Dean Edmonds built SNARC — the Stochastic Neural Analog Reinforcement Calculator, which was arguably the first neural network machine. It relied on forty Hebb’s synapses and utilized a large mechanical system of clutches and knobs to “learn”. Fast forward nearly 70 years, and the largest neural networks have 16 million synapses (equivalent to the computational power of a frog) and run on hardware built with 7nm transistors. These powerful networks are now used for everything from detecting Alzheimer’s disease to optimizing global supply chains.

With such huge improvements and profound impacts of this technology, the logical conclusion is that corresponding capabilities would have been developed for widespread use. In reality, that is not the case. The tools, techniques, and workflows needed to properly utilize these extraordinarily powerful computational methods are still only available to a small portion of the population. Technology giants such as Google, Amazon, Facebook and Microsoft have spent billions building internal teams and competencies to navigate the complexities of utilizing AI/ML in their businesses. The rest of the economy is unfortunately left with their faces pressed up against the glass.

The core issue with these frameworks is not that any of the individual parts are particularly complicated, but that integrating them for maximum efficiency is unbelievably complex. In order to properly build, train, optimize, and deploy AI algorithms, one needs to understand new hardware architectures, data formats, programming languages, and deployment methodologies. The world has not seen this kind of full-stack requirement for a new technology since the advent of the first microprocessor.

The value of understanding and utilizing these tools has not yet been properly quantified, but it is very, very large. Depending on whom you ask, you’re likely to find numbers between hundreds of billions and trillions of dollars. Beyond the economic value, there are clearly tremendous societal benefits in transportation, commerce, food supply, healthcare, and communication that will be fully realized only after these tools have widespread adoption.

With these parameters in mind, we’re thrilled to announce our investment in Spell. Comprised of an amazing team on a mission to expand access to AI, Spell has built the same ML and deep learning tools and platform that give big companies like Google and Facebook an edge and opened them up to the world. Spell’s platform enables any enterprise to utilize cutting edge hardware and software tools to truly unlock the potential of AI.

We’re excited to embark on this mission with the team to bring these capabilities to millions of users and see what they build. The resulting innovations will likely change the world in ways we haven’t yet imagined.

Follow Eclipse Ventures on LinkedIn and Twitter for the latest on the Industrial Evolution.