Separating Fact From Assumption in Today’s Supply Chain Constraints

The full impact of the current supply‑chain constraints within the technology industry is not yet widely understood, which has led to confusion among customers about how and when they may be affected. With conditions evolving rapidly, time is a critical factor. Over the past six weeks, three statements have consistently surfaced in nearly every discussion as our visibility into the situation has improved and the breadth of these challenges has become clearer.

1. This is a memory shortage; only servers in my data center are affected.
2. We’ve navigated constraints before. I can wait this out.
3. Moving everything to the cloud will protect me from the shortages.

Addressing these misconceptions now is essential to making informed, timely infrastructure decisions before constraints translate into real operational risk.

This Is a Memory Shortage; Only Servers in My Data Center Are Affected.

To be clear, this is a multi‑category supply shortage. While memory is currently the most significantly and deeply impacted component—and its constraints create cascading effects across the stack—every major component required to support AI‑driven data center workloads is experiencing pressure. This includes storage media at all tiers, from performance flash to capacity drives.

Nearly every hardware appliance in a data center contains some amount of RAM, including storage arrays. Modern storage controllers rely heavily on RAM because their operating systems use it to support:

• Metadata caching
• Write journaling and NVRAM operations
• Inline data reduction pipelines (compression and deduplication)
• General read/write caching

As a result, current DRAM constraints directly impact storage arrays by limiting controller configurations, increasing bill‑of‑materials costs and constraining availability.

We’ve Navigated Constraints Before. I Can Wait This One Out.

Unlike the supply‑chain constraints between 2020 and 2022—which were driven largely by pandemic‑related factory shutdowns, logistics bottlenecks and a sudden demand for laptops and home office equipment—today’s constraints are fundamentally different.

The current shortage is the direct result of unprecedented demand for infrastructure capable of supporting AI workloads, consuming global memory and storage supply at a pace the industry has never experienced.

Analysts report that AI data centers are now absorbing the lion’s share of DRAM, NAND flash, HDDs and SSD production, leaving far less capacity for traditional enterprise and OEM buyers.

Manufacturers have also reallocated wafer capacity toward High Bandwidth Memory (HBM) for AI accelerators, which has sharply reduced the availability of conventional DRAM such as DDR4 and DDR5.

As a result, there is a structural supply‑demand imbalance projected to persist for at least 12 to 18 months, with reports showing:

• DRAM contract prices on the rise: DRAM contract prices rising 55–60% quarter over quarter heading into 2026.
• Acceleration of NAND contract prices: NAND contract prices climbing 33–60% month over month through late 2025 into 2026.
• Extended enterprise HDD lead times: Enterprise HDD lead times reaching up to 52 weeks—or even 24 months in some cases—due to AI‑driven demand spikes.

Given this environment, prices will continue to increase, and lead times will continue to extend through 2026.

Industry analysis is unanimous: the shortage is structural, not temporary, and manufacturers are prioritizing AI customers willing to pay premium pricing. Waiting to procure infrastructure will result in longer delays, higher costs and reduced allocation priority—ultimately putting organizations further behind competitors who act expeditiously.

Moving Everything to the Cloud Will Protect Me From the Shortages.

A common misconception is that on‑premises hardware constraints can simply be avoided by shifting workloads to the cloud until supply stabilizes.

In reality, cloud providers are subject to the same global component shortages and cost pressures as server and storage manufacturers. As the costs of critical components—particularly GPUs, DRAM, SSDs and high‑capacity HDDs—continue to rise, cloud infrastructure costs are expected to increase as well.

Cloud providers are already experiencing GPU capacity constraints that limit their ability to meet demand for AI and high‑performance workloads. At the same time, memory and storage cost inflation increases the underlying cost of expanding and operating cloud infrastructure.

While there is not yet a defined timeline for when these cost pressures may be passed down to customers, the trend indicates that price impacts are likely over time.

No segment is insulated from these supply chain limitations—not cloud providers, not hardware manufacturers and not resellers.

CDW’s Capacity Analysis and Reporting Assessment (CARA) is a strategic first step toward a more agile and efficient IT environment. Fully funded by CDW, this no‑cost assessment delivers clear, actionable insights into virtualization best practices and resource optimization across your infrastructure.

CARA provides the information and expert guidance you need to right‑size capacity, optimize workload performance and maximize the value of your servers, appliances, virtual machines and software.

Key highlights of the assessment include:

• Best‑practice benchmarking: Compute and resource utilization targets to guide optimal performance
• Virtualization storage insights: Onboard benchmarking for capacity, peak IOPs and throughput
• Operational observations: Clear readouts on cluster, host and VM capacity and utilization
• OS and capacity forecasting: Concise, actionable guidance for ongoing scalability and performance

Together, these insights deliver a clearer, more strategic path forward—reducing complexity and accelerating decision‑making capabilities. With CARA, you can confidently plan for growth, manage resource availability challenges and navigate future uncertainty.