AI is the most compute, network, and data-intensive workload of our time, and to effectively deliver on its promise, solutions must be hybrid by design and built with modern AI-ready architecture.
Organizations have ramped up their AI investments and adoption, with AI’s market size now expected to reach $738 billion by 2030. However, they often overlook key areas of enterprise AI planning, including necessary data maturity levels, networking and compute provisioning, and vital ethics and compliance considerations.
According to our “Architect an AI Advantage” research report, based on a comprehensive survey of IT leaders across 14 countries, critical disconnects in AI strategy have been uncovered. The report highlights an alarming lack of alignment between strategies, processes, and metrics, further complicating the delivery of AI outcomes.
While the overall findings demonstrate the appetite for AI – with nearly all IT leaders planning to increase their AI spend over the next 12 months – they also highlight very real blind spots that could see progress stagnate if a more holistic approach is not followed. For example, misalignment on strategy and department involvement can prevent organizations from leveraging critical areas of expertise, making effective and efficient decisions, and ensuring AI benefits all areas of the business.
See also: A Strategy for Conquering the Digital Maturity Pyramid
Blind spot 1: Low data maturity
Strong AI results that impact business outcomes obviously depend on quality data input. The report confirms organizations clearly understand this concept. But even though a majority cited data management as one of the most critical elements for AI success, their data maturity levels remain low:
- Only 7% of organizations can run real-time data pushes/pulls to enable innovation and external data monetization
- Only 26% have set up data governance models and can run advanced analytics
Further, fewer than 60% of respondents said their organization is capable of handling any of the key stages of data preparation for use in AI models:
- 59% can access
- 57% can store
- 55% can process
- 51% can recover
This discrepancy risks slowing down the AI model creation process and increases the probability the model will deliver inaccurate insights and a negative ROI.
Equally concerning, only 37% of IT leaders have set up shared data models with centralized business intelligence. This echoes previous findings in HPE’s 2022 survey about inadequate data capabilities, where 34% of respondents said their company’s data was isolated in individual applications or locations. Removing data siloes across hybrid architectures is critical to success, and slow progress on this front is a red flag.
To optimize AI performance, organizations must review their technology setup for enabling AI processes across the lifecycle, which demands consideration of multiple elements from skills resources to software, data management, and more. An overarching data and analytics architecture that consolidates all data across applications and locations is a must. The goal should be to provide unified access to real-time data across the organization, no matter where it lives.
Blind spot 2: Deficiencies in networking and compute provisioning
With the data building blocks in place, organizations must seek to understand AI’s specific networking and compute requirements across its end-to-end lifecycle; here, the report also found some issues emerging.
On the surface, organizations seem confident in this area:
- 93% of IT leaders believe their network infrastructure is set up to support AI traffic
- 84% agree their systems have enough flexibility in compute capacity to support the unique demands across different stages of the AI lifecycle
But less than half of IT leaders surveyed admitted to having a full understanding of what the networking or compute demands might be of the various AI workloads across training, tuning and inferencing, which calls into serious question how accurately they can provision for them.
Blind spot 3: Vital ethics and compliance considerations
The report further revealed organizations have failed to connect the dots between key areas of business, with 28% of IT leaders describing their organization’s overall AI approach as “fragmented.” As proof, 35% of organizations have chosen to create separate AI strategies for individual functions, while 32% are creating different sets of goals altogether.
Equally troubling, a majority completely overlook ethics and compliance despite growing consumer and regulatory scrutiny:
- Only 13% of IT leaders deemed legal/compliance to be critical for AI success
- Just 11% of IT leaders deemed ethics to be critical for AI success
- 22% of organizations don’t involve legal teams at all in their business’s AI strategy conversations
This neglect is a serious blind spot, as ethics and compliance will become increasingly important to consumers, and to meet regulatory compliance, more countries will be introduced going forward. Without proper compliance, organizations risk exposing proprietary data – a cornerstone for retaining a competitive edge and maintaining brand reputation. And if they engineer new products without an effective AI policy, they may develop models that lack proper diversity standards, which could hurt brand reputation, cause sales losses, or lead to costly fines and legal battles.
Solutions and considerations
The report noted that if businesses continue with their current approach to AI, it could adversely impact their long-term success. But there are solutions and strategies to avoid the blind spots in enterprise AI planning.
First, organizations must adopt a comprehensive end-to-end approach across the full AI lifecycle to streamline interoperability and better identify risks and opportunities.
Don’t rush the adoption of AI simply because it’s a trending technology. The AI journey should begin with a list of desired business outcomes and leadership input from across the organization on where AI could best help achieve outlined goals.
Have an overarching AI strategy followed business-wide to ensure everyone is working towards the same goals, keeping all considerations – from ethics to sustainability – top of mind.
Ensure the C-suite and IT leaders work collaboratively on the AI strategy, tapping into the business knowledge of the leadership team and the technical expertise of the IT team.
And finally, demand a nuanced approach based on a greater understanding of the AI lifecycle that includes appropriate provisioning from data and compute to software and networking. With hybrid as the dominant operating model, organizations are well-positioned to optimize their capabilities but may need to leverage external experts if they identify gaps in this knowledge.
AI is the most compute, network, and data-intensive workload of our time, and to effectively deliver on its promise, solutions must be hybrid by design and built with modern AI-ready architecture. However, businesses must carefully weigh the balance of being a first mover and the risk associated with not fully understanding the gaps across the AI lifecycle; otherwise, large capital investments in AI can deliver a negative ROI.
