Implementing artificial intelligence in many organisations was supposed to be like turning on a light – a process that is quick, seamless and instantly brightens the business horizon. The reality, however, is proving to be much more challenging, resembling more like building an entire power plant from scratch. The success of advanced algorithms today does not solely depend on choosing the right model, but more importantly on keeping infrastructure costs in check before the technology starts to earn its keep.
According to Wasabi Technologies’ latest Cloud Storage Index report, investment in artificial intelligence is growing at an exponential rate. What is surprising, however, is that as much as 65 per cent of these budgets do not feed into the accounts of innovative software developers at all, but instead flow broadly towards the foundations: storage, data storage systems and pure computing power.
Valley of disappointment vs. incubation phase
The clash between hype announcements and hard financial data is sometimes painful. Currently, only 29 per cent of surveyed companies in the German market report a positive return on investment in AI-based projects. On the surface, this result could be cause for concern, but a deeper analysis reveals a completely different picture. As many as 62 per cent of organisations assume that these investments will start generating real returns in the next twelve months. This phenomenon can be referred to as deferred ROI.
Businesses are coming to the realisation that implementing artificial intelligence is not a sprint, but an extremely demanding marathon. Analytical models require time, vast amounts of precise information and advanced training. Before the expected productivity gains and new business models can emerge, organisations must endure a long incubation period in which capital is allocated intensively without immediate, measurable financial results.
The cloud bill of horrors and hidden costs
During this transitional period, unforeseen infrastructure costs become the biggest threat to the liquidity of innovation projects. The referenced report exposes an inconvenient truth, indicating that almost 48 per cent of companies exceeded their budgets for cloud services in the past year. The reason for this is rarely a mere physical lack of disk space.
Far more often than not, budgets melt in the clash with hidden fees. Half of the expenditure on cloud storage is often additional costs, related to information transfer, API queries or complex access management. The aggregation and processing of the terabytes of data required to power artificial intelligence models generates gigantic network traffic, for which cloud providers bill heavily.
An additional burden is the poor quality of the data itself. Storing unstructured, duplicated or erroneous information costs twice. First, it generates unnecessary storage costs, and then it leads to useless, error-laden algorithm results, which ultimately nullifies the entire investment effort.
Escape to hybrid architecture
The answer to rising costs and system complexity is the growing popularity of hybrid environments. More than 64 per cent of businesses are choosing to combine a local server infrastructure with a public cloud. This division of roles appears to be the optimal compromise in times of market uncertainty. The public cloud takes on the heaviest tasks of aggregating huge data sets and long-term archiving, representing the beginning and end of the analytics pipeline.
Local servers, on the other hand, are used to securely process the company’s most sensitive, strategic assets. However, it is important to remember that this hybrid solution, while extremely flexible, drastically increases the complexity of managing the entire IT ecosystem. Successful orchestration of such an environment requires outstanding architectural expertise to ensure that the cost of transferring data between different zones does not eat up the gains generated by optimisation alone.
A question of trust in the shadow of cyber attacks
Even the best-optimised infrastructure loses its relevance in the face of security breaches. The problem is extremely acute, given that almost half of the companies surveyed have experienced a cyber attack that affected access to their data stored in the public cloud. This situation creates a deep crisis of trust. A significant proportion of users of such solutions do not have complete confidence that their digital assets remain intact after a security incident.
The business consequences could be catastrophic in this case. If AI-based systems start making strategic financial or operational decisions based on data that has been inadvertently modified by an intruder, the entire organisation will be on the brink of the precipice. Therefore, the stability and absolute security of the data storage architecture is an absolute prerequisite before any implementation of advanced analytics.
The foundations of true innovation
True technological transformation rarely begins with flashy visions spun in boardrooms. Its foundations are poured in carefully designed, secure data centres. Before an organisation decides to purchase expensive artificial intelligence-based software licences, it is essential to conduct a rigorous audit of its existing architecture.
Particular attention must be paid to cost transparency, elimination of hidden fees, rigorous hygiene of collected information and unwavering security of the entire ecosystem. Advanced algorithms are not forgiving of digital clutter. The sooner companies get their technological foundation in order, the more efficiently they will join the elite group of those organisations that can already turn the potential of artificial intelligence into measurable business returns.
