Unleashing the power of the AI factory
In “Competing in the Age of AI, Marco Iansiti and Karim Lakhani, warn that a new breed of company is reshaping the way business is done.
Unlike many traditional businesses, these firms are operating under an “AI factory” model which uses a mix of big data and advanced machine learning algorithms to radically transform a wide range of core business functions. The AI factory model leverages exponential economics which if slow to develop at the outset, yet inevitably accelerate at some point to tilt market dominance.
1. Opportunities for incumbents
The AI factory paradigm shift is still in its infancy and may be of large value even outside first- movers.
Firstly, the components of AI technology and machine learning continue to improve rapidly and may lead to a “second mover advantage.” For example, China’s ByTeDance has adopted the AI factory approach through its flagship product TikTok, which automatically delivers a direct stream of short, personalised videos to users, rather than relying on recommendations. The platform now dominates the short-form video market, despite YouTube’s early leadership.
Second, while digital natives have embraced the new AI factory shift, traditional incumbents have been slow to adopt, giving an opportunities for those daring to change. A small number of companies are working tirelessly to prove that they can become successful AI- centric companies. In the financial space, Spain’s second largest bank, BBVA, has embarked on its AI journey by 2015, with the aim of becoming an AI-centric company. As a result, it has built a mobile banking platform with a factory framework that includes agile prototyping of new AI-based financial products, as well as automated trading and interactive services. John Deere, the US-based agricultural equipment company, has built an extensive AI factory with the aim of providing new value-added services to its customers, supporting preventive maintenance and optimal use of its intelligent equipment. Japan’s Rakuten launched its AI-based factory a few years ago and is now one of the best Catalogue-as-a-Service businesses supporting advertisers and retailers on its marketplace.
2 Inside the AI factory model: five new research-based insights
However, given the promise of the AI factory model, it is crucial to analyse more systematically how many incumbents are preparing to migrate to an AI factory model and whether/why they are (not) on their way.to a successful journey. The research we havev recently done began with the publication of an online survey to which over 20,000 managers clicked and 5800 responded. responded. Given the large dimensions of the data collection, we used lasso techniques to identify the most influential AI elements that drive business performance. The study is global in scope .
We gathered five insights.
It’s not too late to make the shift
Among global companies with an established AI strategy, we see a fairly consistent picture across continents and sectors: on average, only 12% of companies have all the building blocks in place for the AI factory paradigm shift. We call them “AI transformers”. Not surprisingly, the most advanced is high tech, and the laggards are healthcare and chemicals, but industrials, energy, media and life sciences are surprisingly well prepared.
Europe has the same proportion of ‘AI transformers’ as the US, but lags in sectors such as chemicals and high tech.
In general, a large proportion of companies (60%) have only the core elements of the AI factory, primarily the technology platform, but they often fail to integrate operations and data at the enterprise level, and they are already struggling to recruit and retain data science skills. These companies typically already have some AI use cases at scale, but these tend to be functional. The lack of cohesion across the enterprise also results in a piecemeal strategy that does not become the ‘de facto’ corporate strategy going forward.
A big reward
There is an emerging body of literature on AI that provides a systematic, statistical analysis of how AI-using companies improve the economics of their businesses. The range of impacts on productivity and profit increases, has been documented to be between 3–5% per year.
This increase in performance is twice the rate of performance of the first generation of computing technologies. Our survey also finds that AI impact on profits is similar to that in the recent literature and implies attractive returns of north of 35–40%.More importantly, 70% of the profit impact gap observed between companies investing in AI is closely related to t the maturity of the AI factory model. C ompanies that successfully leverage all of thebuilding blocks of the AI factory model, we find that they grow 50% faster than others and have more than twice the return on AI capital than the rest of the business population.
The AI factory is more than its core factory
Much of the focus on AI success has been on master in the right production inputs, e.g. getting the right model, accessing enough data, and hiring enough data scientists. While these are necessary inputs to the AI factory model, they are not sufficient. In fact, we have used machine learning to assess the relative contribution of the building of the AI factory model to profit improvement. Our research suggests that only 50% of the profit impact can be attributed to the core technology elements (e.g. technology platform, ML predictive models, data and models, data and data scientists).
The other half is related to complementary assets and practices, such as proper governance (e.g. for data: Establishing a comprehensive framework for responsible data and AI practices; for talent: partnering at scale between business / domainbusiness/domain experts and ML experts) and leadership (theC-suite receives mandatory AI training and supports AItransformation). These blocks are also interdependent — so governance or leadership has a multiplier effect on th eimpact on data and talent, for example.
The AI factory shift is a broad skills shift
The AI skills shortage is an often-cited challenge. While much has been said about the importance of hiring and retaining a core group of data scientists, the commitment of sufficient data science resources remains a constraint today, but it appears to be more of an issue at the start of the AI journey than in extracting value from the transition to the AI factor.
In fact, only7% of companies embarking on the AI journey cite recruitment as a major bottleneckUsing our machine learning technique, we find, not surprisingly, that the architecture of the technology platform architecture is becoming less relevant (as companies have invested in such a platform), but that a critical aspect of scaling is all employee talents
The value of the AI factory is embedded in innovation
Digital technologies are “strategic” in nature, meaning that their versatility enables companies to create powerful, often disruptive new innovations. The AI factory model is a new operating model that has the capacity to industrialise and create a wide range of new product/service and business model innovations.and business model innovations.
Recent academic research confirms that the power of AI lies in product innovation rather than purely efficient process innovation. Our research also explores the role of innovation practices to complement the AI factory model. Unsurprisingly, innovation does not contribute to efficiency, but it does contribute as much as governance to revenue expansion through AI. This effect is also exponential, doubling when AI’s contribution to revenue exceeds 20% of revenue.
