The newly released Wasabi Cloud Storage Index has uncovered a massive paradigm shift in how modern enterprises fund their technology stacks. In a dramatic departure from traditional cloud market dynamics—where software and SaaS platforms historically commanded the largest share of IT wallets—the global race to deploy artificial intelligence has forced infrastructure services back to the center stage.
According to global survey data from 1,700 IT decision-makers, organizations are heavily prioritizing raw computing power, data pipelines, and scalable storage layers over off-the-shelf software solutions. For tech ecosystems and enterprise frameworks analyzing global scalability, this data reveals both the technical blueprints and the hidden financial traps of the current AI boom.
The AI Budget Flip: Infrastructure Beats Software
Historically, public cloud revenue has been overwhelmingly driven by software-as-a-service (SaaS) models. However, the unique demands of training, tuning, and running large-scale AI models require massive computational baselines and immense data volumes.
The report highlights that approximately two-thirds of AI-dedicated budgets are now funneled into underlying infrastructure (data, storage, and processing compute), leaving only a third for software applications.
Global vs. Japan AI Budget Distributions
The structural shifts are incredibly consistent when comparing global data against localized markets like Japan:
| Metric Evaluated | Japan Cohort | Global Cohort |
| Planning an increase in AI infrastructure spending | 64% | 60% |
| Budget allocated to data, storage, and compute | 67% | 66% |
| Budget allocated to pre-built AI software and SaaS | 33% | 33% |
| Planning to lower overall AI infrastructure budgets | 0% | 3% |
This infrastructure spending represents a foundational buildup. Before companies can deploy consumer-facing AI agents or advanced internal enterprise bots, they must first build the cost-effective data pipelines capable of feeding those models clean, high-quality information.
Hybrid Storage Becomes the Architecture of Choice
Managing massive data pipelines has introduced vast structural complexities. To adapt, the majority of enterprises are turning away from pure public cloud environments, choosing instead to deploy hybrid storage frameworks that mix on-premises infrastructure with public cloud tiers.
The survey indicates that 61% of Japanese organizations and 64% globally utilize hybrid storage to manage their active AI workflows. Interestingly, public cloud storage is increasingly being used to “bookend” the data lifecycle, serving two highly specific stages of the pipeline:
- Ingestion & Aggregation: Pulling raw data from disparate, global sources into a centralized, accessible cloud layer.
- Retention & Archiving: Storing finalized models and historical datasets long-term for future audit or iterative training phases.
Hidden Cost Traps: The Egress and API Fee Problem
While infrastructure demand is skyrocketing, legacy pricing models from hyperscale cloud providers continue to trigger severe financial friction. Complex, unpredictable billing structures are leading to widespread budget overruns.
The 50% Fee Tax: For four consecutive years, the data highlights a systemic industry issue: roughly half of all public cloud storage spending goes toward hidden line items like egress fees (data transfer charges) and API call operations, rather than actual storage capacity.
Compounded by massive data growth, these complex fee structures caused 49% of all surveyed organizations to completely overshoot their cloud storage budgets. In fact, 93% of respondents in Japan pointed directly to fee-related line items as the primary catalyst for their budget overruns. When data movement is penalized financially, scaling an AI model that constantly retrieves and processes data quickly becomes unsustainably expensive.
The Emerging Cloud Security Gap
Beyond cost, data protection remains an urgent vulnerability for engineering and IT leadership. Because AI datasets represent highly concentrated, high-value corporate intellectual property, they have become primary targets for malicious actors.
- Frequent Disruptions: 42% of organizations in Japan and 44% globally reported experiencing a cyberattack that resulted in a total loss of access to their public cloud data.
- The Tooling Gap: Alarmingly, 46% of Japanese respondents and 41% globally stated that their current public cloud vendors do not provide the necessary security features to adequately mitigate these modern cyber risks.
This gap underscores a critical market demand for independent, high-performance object storage solutions that offer built-in data immutability—preventing ransomware from deleting or altering backups—without tying security features behind expensive premium tiers.
As enterprises continue to scale AI applications, the strategic selection of predictable, high-speed, and secure infrastructure will ultimately determine whether an AI project delivers a clear return on investment or dissolves into an unpredictable cloud budget deficit.
