Palo Alto, CA – September 18, 2017

Woodside Capital Partners (WCP) is pleased to announce the newly expanded publication of Artificial Intelligence: The Ultimate Technological Disruption. Artificial Intelligence (AI) software performs complex tasks of learning and cognition at a level that matches or exceeds that of humans. This characteristic makes AI a particularly unique technology from the perspective of business models and value creation, as it simulates (and often exceeds) human performance. The business of AI, while quietly advancing in relative obscurity for decades, has finally emerged from its chrysalis into an entirely new industry with its own multi-billion dollar titans, and will force incumbents to acquire aggressively to maintain competitiveness. It is our projection that the value of M&A and Private Placement transactions in AI over the next 5 years will exceed that of the previous 50 years, with several acquisitions topping the $1 Billion mark.

Artificial Intelligence has been an all-purpose term that has been in use since the late 1950s, indeed even before the Integrated Circuit. During most of that time, AI researchers maintained that AI was to arrive in the near future. Yet it did not appear to be getting any closer, and this period of glacial progress was coined as the ‘AI Winter’. For this reason, skeptics continue to point out that this time may be yet another false dawn. But this time, a number of trends that have been quietly progressing have reached certain thresholds. To see why AI has entered into a new era, we must begin with an examination of the enabling factors of AI.

Technological Factors: As with many sudden disruptions, the emergence of AI is the product of not just a single enabling factor that could be predicted with linear projections, but a combination of multiple enabling factors. Two in particular stand out:

1) Inexpensive Parallel Processing: Traditional computer processors could only process information linearly, but many aspects of human intelligence are of a more parallel nature. For example, in order to understand a word, each syllable has to be assessed in relation to each syllable around it, and then each word within the context of a sentence. To see an image, each pixel has to be seen in relation to the other pixels surrounding it, and only then can the image be recognized. Serial computing cannot tackle these tasks with any efficiency.

Enter the demand for inexpensive parallel computing. Initially, it was Gaming, rather than AI, that was the initial catalyst for a consumer-level parallel computing capability. When the highly visual demands of the video game industry proved too much for regular computer CPUs, this gave rise to parallel processors like Nvidia’s Graphical Processing Unit (GPU). The GPU was originally an ad-on to the PC motherboard, but the size of gaming demand led to millions of such GPUs being produced, from which the cost and accessibility of parallel processing improved. As GPUs became commoditized, it was discovered that the speed at which neural networks (computational systems modeled around the workings of a biological brain) could be processed was 25 to 40 times faster with GPUs relative to serial CPUs. Continued price declines in GPUs will further facilitate the advances in AI, even by individual coders.

2) Big Data: For an AI to learn, it needs a certain number of examples through which to cycle through. An AI learns though an iterative process, and thousands or millions of examples of something (whether photos, music, words in a dictionary) have to be processed before the AI becomes competent in a particular task. The level of data available for an AI to access has a significant impact on the speed of learning, and ultimate competence it can attain. Google, for example, has been delivering more precise results in searches, both of websites and photos, due to the unprecedented volumes of data available to Google’s AI algorithms.

As data volumes have risen greatly, there is now a critical mass of data that has crossed the minimum threshold for AI to feed off of to start a virtuous cycle of recursive improvement.
Ultimately, the combination of these two seemingly unrelated areas, parallel processing and big data, sufficiently lowered the input costs needed for AI to advance rapidly and for viable businesses to form around AI. Even better, both still have quite a long runway of growth ahead of them, so AI is expected to improve quadratically through this. Google has made their Tensorflow library open source, lowering the barrier to entry of algorithm creation further still.
Market Factors: AI suffers from a peculiar form of treatment from the media, where any advancement in AI is often not recognized as such. For the last three decades, whenever a form of AI managed to become a successful product in the marketplace, it was often reclassified into a new industry of its own, and hence no longer considered a part of the AI complex. Search engines, speech recognition, voice recognition, autonomous vehicles, industrial robotics, and high-frequency trading are examples of this. When the mainstream believed that AI was a fad that had vanished, in reality AI was already everywhere.

To analyze the deeper market trends of AI, it is necessary to examine the manner in which the media has been tracking AI. Historically, AI has always been a moving target, with each milestone converted into a status quo without most observers noticing. For example, at one time the seminal threshold in AI advancement was declared as the day when an AI managed to defeat the greatest human chess grandmaster. But when this happened in 1997, it was quickly forgotten. After that, the next officially significant threshold was for an AI to win the game show Jeopardy!, considered a formidable challenge due to the presence of trick questions. Yet this summit was conquered in 2011, and was no longer considered important.

The same may be true for future thresholds, such as the Turing Test, created by Alan Turing in 1950, which is when an AI algorithm can pass an interactive dialog test to a degree that makes it indistinguishable from a human. By the most lenient measures, the Turing Test has already been passed by certain subcategories of AI, since certain chat bots can communicate at parity with humans. If a stricter standard is required, there is a public bet between Ray Kurzweil and Mitch Kapor regarding whether AI will be able to pass the Turing Test through eight hours of interviews by 2029, but if current patterns hold, this will be anticlimactic by that time. This tendency to keep extending the threshold that AI has to surpass in order for a new era to begin has led many people to underestimate the quiet progress in the field of AI, leaving them surprised when the sector does manage to become a large industry in its own right.

In this research report, we will examine i) the broader trends in the AI industry, ii) M&A activity, iii) Private Placement financings, iv) public companies in a leadership position, and v) innovative private companies that are candidates for transactions.

Click here to view the full report.

Press Inquiries:
Kartik Gada, Executive Director
Woodside Capital Partners, LLC
[email protected]
(510) 209-8885