Intel (INTC) The Six Five Summit: AI Unleashed 2026 summary

Key challenges in enterprise AI infrastructure

• Enterprises face pressure to move AI from pilots to production, balancing speed, flexibility, and cost-effectiveness.

• Rising token costs and the need to manage ROI are central concerns as workloads scale.

• Not all tokens are equal; efficiency and use-case specificity drive ROI, not just volume.

• Agentic AI introduces new complexities, requiring orchestration and careful workload placement.

• Memory and infrastructure costs are rising, demanding greater efficiency and system optimization.

Evolving infrastructure and optimization strategies

• Hybrid deployments combining on-prem and cloud are emerging as the norm for scalable, cost-effective AI.

• Efficient agent loops, prompt engineering, and observability tools help control costs and improve ROI.

• Open weight models are increasingly viable, allowing enterprises to tailor model selection to workload needs.

• Benchmarking is shifting from tokens per second to business outcomes and correct work per token.

• Infrastructure must balance CPU, GPU, accelerators, storage, and network to optimize end-to-end throughput.

Data, security, and future inflection points

• Enterprises hold critical data that will drive the next wave of AI value creation.

• Protecting proprietary data while leveraging AI requires robust harnesses and hybrid architectures.

• Fine-tuned, domain-specific models can outperform larger, general models for enterprise needs.

• The key mindset shift is to bring AI to the data, not the other way around, to maintain control and maximize value.

• The winners will be those who secure their data and integrate AI efficiently into business processes.