Shortage and the Semiconductor Crisis
Why Hardware Prices and Availability Have Gone Crazy
1. Why Hardware Became Hard to Predict
For many years, buying business hardware was relatively predictable. Prices changed, new generations arrived and individual products occasionally became difficult to source, but companies and IT departments could generally assume that the hardware they needed would be available within a reasonable time and at a reasonably predictable price.
That assumption can no longer be taken for granted.
The current situation is not caused by one single shortage. Several major developments are happening at the same time.
AI and the rapid expansion of datacenters are creating enormous additional demand for processors, accelerators, memory, storage, networking equipment and power infrastructure. Cloud infrastructure continues to grow, while electrification, electric mobility, renewable energy, industrial automation and other technological transformations compete for many of the same components and raw materials.
At the same time, geopolitical tensions and wars have exposed dependencies throughout global supply chains. Critical materials, rare earths, specialty gases, chemicals and other resources required by modern electronics are extracted or processed in a relatively small number of countries.
The supply chain starts long before the semiconductor factory.
A server assembled in Europe may depend on raw materials processed in Asia, semiconductor manufacturing somewhere else, memory from another producer, components from several countries and highly specialized manufacturing equipment that itself has only a handful of suppliers.
Production Is More Concentrated Than the Number of Brands Suggests
Walk through an IT catalogue and it appears that there are hundreds of manufacturers competing with each other. Look deeper into the supply chain and the picture changes considerably.
Important areas of advanced semiconductor manufacturing are concentrated among very few companies. Memory, NAND flash and hard-drive production are similarly concentrated markets. China is increasingly developing additional semiconductor and memory capacity, but globally many critical technologies still depend on a surprisingly small number of producers and production regions. There is just 3 suppliers producing server RAMs. And the 2nm Chips are produced just by a single vendor. The hard disk market has also just 2-3 producers.
This creates enormous efficiency when everything works. It also creates enormous dependency when it doesn’t. And capacity cannot simply be switched on when demand suddenly increases.
Demand Changes in Months. Production Capacity Changes in Years.
A modern semiconductor fab is one of the most complex industrial facilities in the world. Building new capacity requires enormous investment, specialized equipment, infrastructure, qualified personnel and extensive qualification before meaningful production can begin.
From planning to significant production output, expansion can take years. Even then, a new factory does not eliminate dependencies on raw materials, specialty chemicals, manufacturing equipment, packaging and other parts of the supply chain. This creates a fundamental mismatch.
AI demand, investment cycles and geopolitical events can change the market within months. The industrial capacity required to compensate may take years to build.
And this is no longer visible only in industry reports.
It has reached the buyer.
Servers that once had predictable delivery times may now require months. Certain spare parts can require many months to source. SSDs, memory and other components may be difficult to obtain in the required configuration. Products that are technically still available may be offered at prices far beyond previous expectations.
Inventory itself has become expensive. Distributors and resellers cannot simply keep every possible configuration and spare part in stock while acquisition prices are high and future pricing remains uncertain.
And shortage propagates through the market: when new hardware becomes expensive or unavailable, demand moves toward previous generations, spare parts and refurbished equipment, affecting their prices and availability as well.
The situation can sometimes be summarized rather simply:
The hardware isn’t available. And when it is, it may arrive much later and cost considerably more.
Nobody has a crystal ball for how long individual shortages will last. But the causes behind them cannot necessarily disappear as quickly as they appeared.
2. What This Means for Your Hardware Lifecycle
“Hardware lifecycle management” may sound like something only large datacenters need to worry about.
In reality, every business already has a hardware lifecycle.
You hire a new employee and need a notebook and mobile phone. You open a new branch and need workstations, switches, Wi-Fi and a firewall. A server becomes older and its maintenance contract needs renewal. Storage needs additional capacity. A notebook breaks. A switch fails. Eventually equipment reaches the point where maintaining it no longer makes economic or technical sense and it has to be replaced.
Your hardware lifecycle is simply how you plan, purchase, maintain, support, extend and eventually replace the equipment your business needs.
Until recently, businesses generally followed variations of two models.
The Structured Lifecycle
Larger and more structured organizations usually standardized their infrastructure.
Perhaps notebooks came from one manufacturer, servers from another and networking from another. Hardware was purchased through established channels, covered by maintenance contracts where necessary and replaced according to defined lifecycle periods.
This provided something extremely valuable: predictability.
Budgets could be planned. Replacements could be scheduled. Support contracts could provide defined response and replacement conditions. Standardization made infrastructure easier to operate. That model remains valid where its assumptions continue to work.
But even a well-structured support contract cannot manufacture a component that is unavailable.
There have already been situations where required replacement hardware or spare parts could not be supplied within the expected contractual timeframe because of circumstances outside the normal control of the supplier.
Contracts normally define what happens in such situations. Depending on their terms and circumstances, this may involve service credits, compensation, refunds or exceptional provisions such as force majeure.
Commercially, that matters. Operationally, however:
Compensation for a missed SLA does not make your failed system operational again.
If the replacement part cannot arrive and your production, office, application or infrastructure remains unavailable, receiving contractual compensation solves a different problem.
This means that even organizations with professional lifecycle management may need to reconsider whether the contract alone still provides sufficient operational protection for every critical system.
“If It Breaks, We’ll Buy Another One.”
The second model was much simpler — and for many businesses it was perfectly rational. Why maintain spare hardware or pay for expensive support on equipment that was readily available?
- If a switch broke, order another one.
- If a workstation failed, buy another workstation.
- If a server reached the end of its useful life, order the replacement.
When hardware was broadly available and prices relatively stable, availability itself acted as part of the continuity strategy.
- Today, the local supplier may not have it.
- The distributor may not have it.
- The usual online sources may not have it.
- Or somebody has it — but at a price completely different from the one on which the original replacement strategy was based.
Both traditional models therefore face the same changed market from different directions.
The structured organization may discover that even a contract cannot guarantee physical availability under every circumstance.
The business relying on immediate replacement may discover that immediate replacement is no longer immediate.
Has your hardware lifecycle adapted to this new reality — or is it still based on the availability and prices we used to take for granted?
3. The Answer May Not Be One Model
There is no universal replacement for the traditional hardware lifecycle.
The right approach depends on what the equipment does, how critical it is, what downtime costs, what support is required, what alternatives exist and what they cost. For many businesses, the answer will increasingly be a combination of different approaches.
Keep What Still Works
An existing vendor relationship does not need to disappear because the market has changed. Where manufacturer contracts continue to provide appropriate pricing, delivery times and support conditions, they may remain the best option. The objective is not to replace a functioning model. It is to identify where that model no longer provides the predictability the business requires.
Use More Than One Vendor Where It Makes Sense
Standardization has enormous operational advantages. But absolute dependence on a single manufacturer or sourcing channel can also become a constraint when availability and prices diverge significantly. Different manufacturers can therefore be considered for different parts of the infrastructure, while availability and pricing become part of lifecycle planning rather than something checked only when the purchase order is ready. This does not mean creating an uncontrolled collection of unrelated hardware.
Standardize where standardization creates value. Diversify where dependency creates unnecessary risk.
Keep Selected Hardware Available Before You Need It
When availability itself becomes uncertain, inventory can become part of continuity planning again. That does not mean filling a warehouse with expensive equipment. It means identifying components where the cost of keeping a replacement available is reasonable compared with the impact of not having one.
Sometimes, at today’s prices, keeping a compatible replacement switch on the shelf can be cheaper than accepting the risk of trying to find one after the failure.
The same principle can apply to selected power supplies, controllers, drives, memory, network equipment or even complete systems.
Refurbished Does Not Have to Mean Unsupported
Professionally refurbished enterprise hardware is another legitimate sourcing option. This should not be confused with simply buying an unknown second-hand device from the cheapest online seller. Enterprise refurbishers can provide tested systems and components with warranty, replacement options and, where required, maintenance contracts and defined service levels.
This can make previous hardware generations part of a predictable lifecycle rather than merely an emergency compromise.
Previous Generations May Change the Calculation
Under normal market conditions, choosing older hardware to save a relatively small amount often makes little sense. Newer generations may provide higher performance, lower energy consumption and a longer remaining lifecycle. But the calculation changes when the price difference is no longer small. A previous-generation server may consume more electricity and still be economically preferable if the difference in acquisition cost is large enough. Over three or five years, the additional power consumption may cost less than the premium required for the newer system.
Yesterday’s hardware strategy was calculated with yesterday’s hardware prices. Today’s numbers may produce a different answer.
Manufacturer End of Support Does Not Necessarily Mean End of Life
Planned replacement cycles can also be reconsidered. Manufacturers eventually end support for older product generations. Traditionally, this often became the natural moment to replace the asset.
But specialized post-vendor maintenance providers can continue supporting enterprise servers, storage, networking and other infrastructure beyond the manufacturer’s own lifecycle.
Depending on the equipment, contracts can provide access to spare parts, replacement services and defined service levels.
A system that still provides the required performance may therefore remain operational for additional years without simply accepting an unsupported infrastructure risk. This can be particularly relevant when replacing it during a period of high prices and uncertain availability would provide little immediate business benefit.
Sometimes You Don’t Need to Buy the Hardware at All
Another response is to transfer part of the hardware lifecycle outside the business.
Cloud infrastructure is one possibility, but outsourcing infrastructure does not automatically have to mean moving everything to a global hyperscaler.
Colocation, European infrastructure providers and trusted managed providers can offer alternatives where physical hardware, maintenance and replacement are handled professionally while the customer retains substantially more choice over architecture, location, backups and operational dependencies.
The same principle can apply to smaller environments. A business does not necessarily need to own every server or maintain its own server room simply to retain appropriate control over its systems and data.
The decision should again be based on requirements rather than ideology:
Which infrastructure model provides the appropriate balance between cost, performance, availability, control and dependency risk?
The result may be mixed.
New vendor-supported hardware in one area. Post-vendor maintenance somewhere else. Professionally refurbished equipment for another requirement. Selected spares on the shelf. More than one manufacturer. Some infrastructure owned, some colocated and some provided as a managed service.
That can provide more resilience than forcing every asset into the same lifecycle model.
But it introduces another problem.
Someone has to manage all that complexity.
4. Managing the New Complexity
The traditional hardware model was attractive not only because hardware was predictable. It was also relatively simple to manage. One or a few manufacturers. Established distributors. Defined support contracts. Standard replacement cycles.
The alternatives described above can improve resilience and economics, but they also require more knowledge and coordination. Someone has to understand which systems can safely remain in service longer, which spare parts should be available locally, which refurbished equipment is trustworthy, which maintenance provider can support an older platform, which alternative manufacturer is suitable and where a difficult component may still be available.
Someone also has to have and maintain the relationships required to do it.
For an internal IT manager, managing multiple manufacturers, distributors, refurbishers, post-vendor maintenance providers, infrastructure providers and sourcing channels can itself become a significant workload.
Hardware Lifecycle Management Can Be Outsourced Too
This is where we believe the lifecycle model itself can evolve. You do not necessarily need to replace your internal IT department, your trusted hardware supplier or your existing manufacturer contracts.
We can work alongside them.
- Where an existing vendor contract works, keep it.
- Where another manufacturer provides better availability for a particular requirement, evaluate it.
- Where an existing system can economically remain in service through post-vendor maintenance, extend it.
- Where refurbished enterprise equipment provides the right combination of cost, warranty and support, consider it.
- Where availability creates a continuity risk, identify what should be kept ready in advance.
- And where sourcing becomes difficult, use a broader supplier and partner network instead of relying on a single catalogue.
Think of It a Little Like an Insurance Broker
A company does not necessarily want to maintain relationships with every insurance provider simply because different insurers may provide the best solution for different risks.
A broker knows the market, understands the requirements and helps coordinate the appropriate solutions.
Hardware lifecycle management can work in a similar way.
We already work across different manufacturers, distributors, specialized enterprise refurbishers, maintenance providers and infrastructure partners. We combine that network with technical experience to evaluate not only where something can be purchased, but whether the proposed alternative makes technical and economic sense.
Sometimes the requirement is simple: find a component that has become difficult to obtain. Sometimes it means evaluating several sourcing alternatives. And sometimes the complete hardware lifecycle — support contracts, sourcing, replacement planning, strategic spares and different providers — can itself become a managed function. The customer retains visibility and control over the decisions.
We manage the complexity around them.
An important part of that capability is the network behind it. Over decades of working with enterprise IT, we have built relationships with manufacturers, distributors, specialized refurbishers, maintenance companies and infrastructure providers across Europe and beyond.
Recreating and maintaining that network internally would make little sense for most organizations. For us, it is part of our daily business. The same relationships and sourcing channels we use for our own infrastructure can be used across multiple customer environments — turning individual relationships into a shared capability.
This also means that lifecycle management does not have to depend on a single catalogue, distributor, manufacturer or maintenance provider. When one channel cannot provide the required solution, we can evaluate others.
Our role is not to sell you whatever happens to be available from one supplier. It is to find and coordinate the option that best fits the requirement.
There is no magic solution to a global semiconductor shortage. We cannot make a component appear that physically does not exist.
But technical knowledge, multiple sourcing channels, maintenance alternatives, planning and a strong network can substantially reduce the number of situations in which one unavailable component becomes an unpredictable business problem.
Keep what works.
Extend what still makes sense.
Replace what needs to change.
Prepare for what may become difficult to replace.
And when something has already become difficult to find:
Put us to the test.
Talk to us about your hardware lifecycle →