As seen in a previous article, despite being at the forefront of fundamental AI research spurring the latest innovation breakthroughs in Machine Learning (ML), Canada’s bid to build domestic AI champions is increasingly challenged by established tech giants. Indeed, bearing in mind that AI expertise is one of the key resources required to become an AI power, Canada is facing a brain drain of its precious talents, often recruited by global tech players.
Canadian authorities want to entice AI talents to stay and build the domestic AI industry. To do so, they devised a strategy to provide early and sustained funding to grow the AI talent pool and finance its ability to not only further Canada’s lead in fundamental research undertaken in academic institutes, but also conduct applied AI research within startups generating incremental innovations consisting in commercialized AI solutions. This public funding in fundamental and applied AI research is mostly allocated to securing computing hardware required to crunch the large volumes of data necessary to train ML algorithms.
In the case of applied research in startups, initial and sustained government funding would attract complementary private funding including seed and VC investments aimed at obtaining computing hardware. Plus, it would entice organizations interested in adopting AI to provide those startups with large volumes of their proprietary data to train ML algorithms suited to their needs. Ultimately, this government strategy aims to get Canadian AI startups to scale up and become world-leading AI companies competing with tech incumbents.
However, with regards to applied AI innovation in startups, this generous public funding isn’t enough to encourage organizations to provide proprietary data to train ML algorithms. And it’s certainly insufficient to entice private investors to allocate complementary funding to secure computing hardware. Tech founders must also have a sound business plan where they identify market segments where they have a shot against Big Tech. Only then can founders attract private investors and entities willing to provide proprietary data for ML algorithms. For that matter, in addition to the governmental strategy, the Canadian tech community needs a game plan that would work in synergy with the former. As such, the tech community’s strategy considers limitations inherent to the Canadian tech ecosystem.
First, it’s very difficult for tech entrepreneurs to develop ambitious ML solutions with a general focus due to the larger and costlier computing infrastructure required. Given the smaller pool of VC in Canada compared to the US, private investors are reluctant to pour money in a project that is too ambitious. Even if VC money was allocated to such a project at the seed and early stages, once it reached mid-size, private investors won’t be able to continue contributing ever larger amounts at each financing round.
Therefore, to continue to develop its ambitious AI solution, the startup has no other choice but to be bought by a Big Tech firm that has the financial resources, computing infrastructure and data to perfect the ML algorithm. As such, this defeats the Canadian authorities’ objective of fostering a Canadian AI champion. Such is the case of the Montreal-based startup Maluuba currently focusing on Artificial General Intelligence (AGI) by developing a ML-powered virtual assistant with a cross-domain expertise that can think and communicate like humans. To continue its project, Maluuba had to agree to a M/A by Microsoft who is providing the computing infrastructure and the vast volumes of data required to train such an AGI algorithm.
Second, the dominance of Big Tech afflicts even Canadian AI startups looking to field narrow-focused AI solutions designed for one expertise in a single domain in the internet consumer market (or B2C) where tech incumbents are jealously defending their foothold. Indeed, since Big Tech amasses and hoards the greatest share of consumer data, significant entrance barriers to the B2C market exist for prospective Canadian AI startups. Those hurdles are part of Big Tech’s arsenal to limit competition coming from any upstart firm looking to break into the consumer market. Ultimately, to obtain this data necessary to train a ML algorithm, a Canadian tech founder with a promising B2C narrow AI solution might have to integrate with a Big Tech firm interested in adopting this software to further cement its supremacy.
Since Big Tech’s superiority in computing infrastructure and consumer data stifles Canadian attempts to independently develop ambitious ground-breaking AI programs and narrow-focused AI B2C solutions, local tech entrepreneurs must concentrate on narrow AI applications in the internet business market by developing enterprise ML software. Indeed, the B2B (and B2G) market hasn’t been cornered by Big Tech firms who tend to underserve businesses by selling them streamlined software that don’t fully satisfy these organizations’ precise needs.
Indeed, since a lot of those businesses’ needs are so specific to one industry, it won’t be profitable for Big Tech to develop a specialized program tailored to one niche, unlike a simple consumer-oriented software which given its larger customer base yields more profits. As a result, Big Tech’s B2B approach is to sell the simplest software possible that can be used by the biggest number of people. Therefore, given that the Big Tech competition is less intense in this space and businesses are still lacking specialized software, prospective Canadian AI startups can develop enterprise AI software tailored to an industry’s particular needs that has the potential to boost productivity.
Since data is the new fuel and a lot of it is required to program ML algorithms, ML-based startups will focus on hatching large domestic and foreign businesses that generate and possess a lot of corporate data. Furthermore, to beat the competition, startups must be willing to provide their services at lower prices than established tech firms. Plus, if possible, association with Canada’s ML pioneers and AI academic institutes is a great marketing strategy to attract traditional corporate companies.
By concentrating on the B2B enterprise software market where there is room for new providers – and therefore real growth prospects – it is hoped that Canadian AI talents will be enticed to stay and build up Canadian champions focused on AI enterprise software, going toe-to-toe with Big Tech.
This synergy between the Canadian authorities’ strategy and the local tech community’s game plan has yielded positive results. With Canadian AI startups focused on B2B solutions, companies are willing to provide data to train ML algorithms. One of the greatest successes is the creation of the Quebec-based Institute for the Valorization of Data (IVADO) gathering industries and AI researchers where the former provides data to train algorithms.
As for private sector funding, between 2017 and 2018, there was a 28% increase in the number of AI startups across Canada bringing the total count to 650 startups country-wide. Regarding private investments, an average of 49% in Startup deals has been registered over the last five years. Furthermore,
Currently, according to the early-stage VC fund 500 Startup, the most popular AI applications pursued by Canadian startups are cyber security, fintech (real estate, personal finance) and business analytics. Furthermore, one should note that Canadian tech entrepreneurs are at the forefront in the development of medical AI solutions including AI-enabled diagnoses. Indeed, being a welfare State, Canada has accumulated and centralized volumes of medical data that is easily accessible for entrepreneurs to use for training healthcare-related ML algorithms.
Thanks to this public/private partnership, the brain drain slowed down and today Canada boasts the third largest AI talent pool in the world giving the country sufficient expertise to create AI champions able to compete with tech incumbents.
However, will this concerted public/private effort be enough to reduce the threat of Big Tech looming over Canada’s bid to become an AI industry leader? A future installment will explore this question.
Photo: Artificial Intelligence (2017) by SeanBatty via Pixabay. Public domain.
Disclaimer: Any views or opinions expressed in articles are solely those of the authors
and do not necessarily represent the views of the NATO Association of Canada.
