X
99851

Update: 2020 Trends in Data, AI & Analytics

July 1 2020
by Matt Aslett, Nick Patience, Krishna Roy, James Curtis, Paige Bartley, Csilla Zsigri, Jeremy Korn


Introduction


The coronavirus outbreak has rapidly shifted global markets, including those related to enterprise technology, since we published our 2020 Trends report on Data, AI and Analytics in November 2019. This report serves as an update to our 2020 predictions in light of market changes due to COVID-19.

The 451 Take

The coronavirus pandemic is having an impact on all sectors of the technology industry. Data, AI and analytics is no exception. That said, there are reasons to believe that this sector will be relatively resilient. There's a reason why respondents to our Digital Pulse survey rated data and analytics as the highest-priority technology for both 2018 and 2019, while data and analytics and machine learning/AI were respectively rated as the number one and two technologies with the greatest game-changing potential in 2020. While those surveys were conducted pre-COVID-19, the results illustrate that data and analytics have the power to drive transformational change. While spending on new analytics projects will potentially be delayed or limited in the short term, enterprises and governments are turning to this field to understand the impact of the current disruption on employees, customers and supply chain partners alike. Even as companies in the industries most affected by coronavirus cut back on non-essential spending with a view to staying afloat, the continued use of data and analytics is arguably more valuable than ever in shaping the strategic decision-making that will ensure their long-term survival.

451 Research's Q4 view of the market – what we said


Figure 1
2020 Trends in Data, AI & Analytics, as of November 2019
451 Research, 2019


Our updated evaluation of each trend follows.

AI infrastructure: still critical, but investments will shift


Given the larger macroeconomic trends of the COVID-19 pandemic, there are logical reasons to expect enterprises might refocus their AI efforts to more immediately pressing use cases (e.g., supply chain optimization or IT network optimization). But because machine learning is a general-purpose technology that enables enterprises to optimize and automate a variety of processes, investment in AI is still a good bet in spite of – and perhaps because of – an economic downturn.

We expect infrastructure, as an integral component of AI, to receive continued focus from enterprise adopters of the technology. It still holds true that the success or failure of an AI initiative is partially dependent on enterprise infrastructure.

That being said, the new imperative of remote work could impact enterprise decision-making about how to distribute investments in AI infrastructure. The decentralization of data science teams outside of office environments could diminish the importance of on-premises infrastructure for AI workloads, such as on-premises servers or networking technology. Instead, organizations might increase their spend on cloud-based AI platforms that both provide a collaborative workspace for disparately located teams and abstract away the complexity of defining and managing infrastructure resources. Over the past year, we've already seen an uptick in interest around cloud-based AI platforms, but the new normal could accelerate this trend even further.

Enterprise intelligence platforms: on an extended timeline


We have been talking for some time about the evolution of the Enterprise Intelligence Platform, designed to consolidate data ingestion/integration, database management and analytics functionality in a single product, and stated that we believed 2020 would mark the beginning of its emergence as a new product category in the analytics space. While the trends driving the consolidation of data ingestion/integration, database management and analytics functionality remain relevant in a 'post-coronavirus' world, the timeline for the evolution of products and services that meet the description will inevitably be impacted.

From a vendor perspective, the evolution of products that could be described as Enterprise Intelligence Platforms was going to be driven by M&A and consolidated product development efforts, both of which are likely to be delayed by the need to address more short-term initiatives. From a customer perspective, adoption of Enterprise Intelligence Platform products and services was likely to be driven by greenfield and brownfield analytics projects. While some enterprises will be accelerating these projects as part of larger transformation acceleration efforts, others will be focused on making do and squeezing as much value as they can from the analytics resources they already have at their disposal.

Automated decision intelligence: taking a back seat to already-proven approaches, but undergoing refinement


Automated decision intelligence will still become a bedrock for business decision-making, but not as quickly as before the coronavirus took hold. The pandemic has slowed the pace of innovation for many software vendors in the analytics software sector and is also likely to have a detrimental effect on buying behavior. Rather than invest in new and yet to be fully proven approaches, of which automated decision intelligence is one, most organizations are likely to double down on conventional techniques that they know users are comfortable with – and that, perhaps more importantly, comes with a defined and measured ROI.

Furthermore, any negative feedback from early adopters will filter through to the mainstream, who are likely to take more notice of it when budgets are tight and businesses have been disrupted. For instance, criticism will be taken on board that natural language generation, if too repetitive and obvious, suffers from wallpaper syndrome and is no longer noticed or valued. Observations that some 'smart' recommendations and suggestions are not so smart as they are all too obvious and automatically generate correlations, which are coincidental rather than casual, will be also taken more seriously. However, this could prompt vendors to refine their automated decision intelligence technology to make it more effective. It could also result in more organizations realizing that augmented features are designed to assist human decision-making and not replace it, which is another positive benefit.

MLOps: short timelines for new projects could expedite decisions


As we wrote in the original report, both enterprise adopters of AI and vendors in this space are still formalizing procedures and standards for the operationalization of AI models. This long-term process will hardly be deterred by effects of the coronavirus pandemic. AI will continue to be important, as will its operationalization.

In theory, a push to fast-track AI initiatives would cause organizations to expedite formalization of MLOps or lead vendors to roll out products and services more quickly than originally planned. On the other hand, a rush to deploy in production without considering MLOps carefully could result in failure of the project. And if, as we outlined in the above section on AI infrastructure, the economic downturn causes enterprises to pivot to different AI strategies based on different underlying infrastructure, then the coronavirus pandemic could perhaps have a demonstrable impact on what standards around MLOps arise.

We have seen evidence of fast-paced AI deployments when it comes to use cases related to the pandemic. A number of AI technology and healthcare providers released coronavirus-related projects – from natural language analysis of academic papers to hospital triage applications. The quick timelines of these projects probably forced technologists to make expedient choices about the operationalization of these projects, whether it be deployment venues, security practices or quality assurance, and these on-the-fly decisions could influence the development of operationalization protocols and best practices, specifically in the healthcare space.

These musings are all fairly hypothetical. In the short term, MLOps will continue to be an important consideration as AI adoption matures in the age of coronavirus.

Data catalogs and metadata management: offerings that enable remote work to rise to the top


As organizations adjust their strategies and technologies to cope with the global impact of COVID-19, metadata management retains its importance in supporting analytics and governance initiatives. Organizations, now as much as ever, are looking to leverage analytics to pinpoint areas where increased operational efficiencies can be realized. What might be temporarily dampened, however, is the long-term vision of the data catalog's critical role in supporting DataOps.

Questions linger as to whether organizations will be willing to invest in new metadata management technology in a time when quick fixes are sought. While 34% of respondents in 451 Research's Digital Pulse, Coronavirus Flash Survey March 2020 report their organization expected to spend more on IT resources/assets as a result of the coronavirus situation, it is likely this spend is largely associated more with alleviating the sudden strains of a remote workforce.

For organizations that have multiple existing data catalog investments, increased and perhaps enduring prevalence of work-from-home may in fact cause organizations to start to favor the offerings that are more compatible with the remote work model. The 'winners' will likely be those with cloud deployment, browser-based access and robust security/authentication that will help workers continue to reap the productivity benefits of catalogs while out of the physical office.

Databases and distributed data processing: timeline for shifting to cloud accelerates


Databases have been a staple of enterprise IT architecture for decades. Clearly, we don't expect that to change because of the pandemic. Enterprises will continue to generate the same or even greater amounts of data during this time and will need database technologies regardless of the pandemic. However, the way enterprises consume their database and data platform technologies could see an adjustment.

Our ongoing research continues to suggest that enterprises will gravitate to cloud computing for their database processing, moving away from an on-premises deployments. Efforts in procuring and installing hardware, as well as the ongoing managing of that hardware infrastructure within a company's datacenter, could potentially run counter to local, county or country health and safety guidelines to maintain proper social distancing. While enterprises already were well on their way to adopt cloud computing and cloud-native technologies for their database needs, the COVID-19 pandemic may speed up that process.

Additionally, the need to improve trust in data and other assets, remove long-standing inefficiencies from business interactions and processes and boost the resilience of supply chains has become more important than ever amid the coronavirus pandemic. These are areas where distributed ledger technologies can be marshalled – alongside analytics, AI and IoT – to improve the overall digital experience.