Storage Crisis, Hyperscalers, Switzerland: What's True – and What's (Still) Storytelling?

Blog (EN)

The starting point is the Clarus News post "Storage Crisis in Europe: Hyperscalers Profit, European Data Centers Fight for Survival" (December 20, 2025). Here's the link:
➡️ https://clarus.news/de/post/speicherkrise-in-europa-hyperscaler-profitieren-europaeische-rechenzentren-kaempfen-ums-ueberleben-20251220 :contentReference[oaicite:0]{index=0}

The article is based on a heise article (interview/statement from a DC specialist) and constructs a very far-reaching diagnosis from it. Below, I examine the hard claims (prices, market mechanics) and separate them from interpretation/prognosis (sovereignty, "starvation death", policy failure). Additionally: Swiss perspective (Swisscom, Green, energy/location issues, data protection).

1) Fact-checking the central claims

Claim A: "Hardware prices ... increased fivefold within four months"

Status: not reliably proven (in this form rather exaggerated / imprecise).

  • The heise text clearly mentions (in the visible part) strongly increased prices in the last four months, but the specific number "5×" is not clearly proven there. :contentReference[oaicite:1]{index=1}
  • What external market signals show for 2025: significantly rising DRAM prices, but depending on segment more like +50% (server DRAM, sometimes acute) or up to ~+170% (contract prices over a year) – so painful, but typically not blanket "5×". :contentReference[oaicite:2]{index=2}
  • Individual end-customer/retail reports mention very strong spikes (e.g., "RAM even up 200%"), which is more product/market-specific than "data center hardware overall". :contentReference[oaicite:3]{index=3}

Criticism: The Clarus text turns "strongly increased" into a precise "5×" metric – without properly defining:

  • Which hardware? DRAM? HBM? RDIMM? SSD? complete servers?
  • Which market? Spot vs. contract, Europe vs. global, hyperscaler vs. channel.

➡️ Better formulated: "Certain storage/server components became drastically more expensive in 2025 within a short time; depending on segment, +50% to multiple times higher prices are reported."

Claim B: "Million-dollar projects are being stopped, tenders withdrawn, legal disputes threaten"

Status: plausible, but anecdotal/qualitative – not proven as comprehensive statistics.

  • The heise interview explicitly mentions threatening legal disputes (price guarantees vs. delivery/prices) and that projects may be postponed because the price increase is "only" a few months old. :contentReference[oaicite:4]{index=4}
  • This is a strong signal, but it remains assessment knowledge from a practitioner – not hard market statistics.

➡️ Takeaway: "Yes, the scenario is realistic and observable in the market – but the Clarus text oversells it as a universally valid situation assessment."

Claim C: "Hyperscalers can financially weather the crisis and expand dominance"

Status: logically comprehensible, argued this way in the heise text – but causality remains assumption.

  • Heise clearly describes the thesis: hyperscalers have "large financial resources" and calculate with time horizons until smaller competitors "starve". :contentReference[oaicite:5]{index=5}
  • At the same time, it's not claimed there that hyperscalers "fabricated" the situation; rather: they are "not sad about it". :contentReference[oaicite:6]{index=6}

➡️ Criticism: The Clarus text presents a very strong competitive story ("calculated market consolidation"). This would require evidence like:

  • Capacity commitments (long-term supply agreements),
  • Displacement effects in supply chains,
  • Price-setting power that demonstrably excludes smaller providers.

What's proven is rather: Scarcity + scale advantages = hyperscalers get through better. :contentReference[oaicite:7]{index=7}

Claim D: "Used market: no alternative – higher power costs, lower performance"

Status: plausible as HPC/high-performance perspective, but not universal.

  • Heise (Daniel Menzel) argues exactly this way: used equipment often means more power costs and less "workload per watt/franc". :contentReference[oaicite:8]{index=8}
  • For certain workloads (storage tiers, less dense compute applications), "used" can still be economical – so the claim is context-dependent.

2) Switzerland connection: Are Swiss DCs "fighting for survival"?

The Swiss reality appears differently weighted: more expansion and acceptance/energy issues than "starvation death"

  • SRF describes a boom of new server farms in 2025 (also AI-driven) and a debate about power consumption (projections until 2030). :contentReference[oaicite:9]{index=9}
  • Swiss operators like Green communicate expansion (e.g., Zurich-West Campus/extension). :contentReference[oaicite:10]{index=10}

➡️ Interpretation: In Switzerland, the bottleneck/crisis story is more:

  • Energy & permits & network connection
  • Location attractiveness for hyperscalers
  • Sovereignty/jurisdiction (US Cloud Act, exit strategies)
    ... and less "DCs dying en masse".

3) How specifically affected are Swisscom and Green?

Swisscom: potentially "dual-track" affected – and partly even in advantageous position

What Swisscom does (relevant to the topic):

  • Swisscom operates data centers/colocation in Switzerland (multiple locations) and positions "data retention in Switzerland". :contentReference[oaicite:11]{index=11}
  • At the same time, Swisscom expands its offering as local partner for hyperscalers (AWS/Azure services, hybrid integration). :contentReference[oaicite:12]{index=12}

What the "storage crisis" can mean for Swisscom:

  • Negative: If Swisscom procures hardware for its own private/managed cloud platforms, DRAM/SSD price jumps hit margins or capex planning. (General market situation: rising DRAM prices) :contentReference[oaicite:13]{index=13}
  • Neutral to positive: As integrator/reseller/managed provider, Swisscom can partially pass on price pressure or compensate through service components (depends on contract models).
  • Potentially positive: In phases of scarcity, customers tend toward "large, stable supply chains" + hybrid setups. Swisscom sits exactly in this intermediate space (local + hyperscaler). :contentReference[oaicite:14]{index=14}

Sovereignty angle: Swisscom actively plays this theme ("digital sovereignty", "data under Swiss sovereignty"). :contentReference[oaicite:15]{index=15}

Green: rather "beneficiary of demand", but indirectly exposed through customer capex and energy

What Green communicates:

  • Green explicitly positions itself as a datacenter provider for hyperscale requirements (multi-megawatt campus) and is expanding. :contentReference[oaicite:16]{index=16}

How the storage crisis can hit Green:

  • Directly: Green primarily sells space/power/cooling/connectivity. DRAM prices are not their core business – the direct cost driver is more energy and construction/technology.
  • Indirectly: If customers (clouds, enterprises) postpone projects due to hardware prices, this can influence rollout timelines (when racks are actually occupied).
  • Counter-cyclical: AI and cloud demand continues to drive the need for data center services; additional hyperscaler investments in Switzerland are real (e.g., Microsoft invests/expands local DC capacities). :contentReference[oaicite:17]{index=17}

Conclusion on Green: The Clarus tone of "fighting for survival" fits Green as market leader/expander only very limitedly. More likely are acceptance/energy/timing risks, not existential struggle.

4) Digital sovereignty: What's the core – and what's often confused?

The Clarus text uses "digital sovereignty" as big slide terms for dependency. This isn't wrong, but too crude.

Practically, sovereignty in Switzerland is usually determined by 5 questions:

  1. Jurisdiction: Under which law can government access take place (e.g., US Cloud Act as concern, even with CH regions)? :contentReference[oaicite:18]{index=18}
  2. Data location & data flows: Where do data really reside (backups, telemetry, support access)?
  3. Control/transparency: Who administers keys, who has root/support access?
  4. Exit capability: Can I get out without operations collapsing (FINMA/banks drive this strongly)? :contentReference[oaicite:19]{index=19}
  5. Technical sovereignty: Standards/open source, portability, multi-cloud/hybrid – instead of lock-in.

Swiss debate 2025: Exactly these points (Cloud Act, governance, exit, transparency) are very present in Swiss IT media. :contentReference[oaicite:20]{index=20}

➡️ Important: A "Swiss cloud region" from a US provider can deliver data location – but doesn't automatically solve jurisdiction and dependency.

5) Data protection Switzerland vs. Europe: the most important differences (practical)

Commonalities

  • The revised Swiss Data Protection Act (revDPA) is closer to GDPR than before; the EU has also confirmed adequate data protection level to Switzerland after revision (important for CH↔EU data flows). :contentReference[oaicite:21]{index=21}

Differences that really matter in cloud/DC questions

  1. Sanctions system

    • Switzerland: Fines up to CHF 250,000, typically against responsible natural persons (criminal law logic). :contentReference[oaicite:22]{index=22}
    • EU (GDPR): up to 20 million EUR or 4% of worldwide turnover, typically against the company. :contentReference[oaicite:23]{index=23}

  2. Supervisory/procedural mechanics

    • EU: "One-Stop-Shop"/Lead Supervisory Authority for cross-border processing (important for multi-EU rollouts). :contentReference[oaicite:24]{index=24}
    • Switzerland: No EU mechanism; primarily FDPIC + cantonal/sector-specific constellations.

  3. Extraterritorial effect

    • GDPR can also affect Swiss companies if they address/track EU persons etc. (Swisscom describes this as practical reality). :contentReference[oaicite:25]{index=25}

➡️ Cloud point: "Swiss data protection" is no free pass – many Swiss companies must master DPA + GDPR simultaneously.

6) What's good about the Clarus article – and what should be refined

Strengths

  • Right direction: Scarcity/price surge for storage/server components is real and AI-driven. :contentReference[oaicite:26]{index=26}
  • Right power analysis: Hyperscalers get through better with capital/scaling. :contentReference[oaicite:27]{index=27}

Weaknesses / Exaggerations

  • The "5× in 4 months" metric appears like a pointed number without proper definition (and is not robustly proven as such). :contentReference[oaicite:28]{index=28}
  • "Europe fights for survival" works as a headline but is too general as diagnosis – Switzerland simultaneously shows expansion and hyperscaler investments. :contentReference[oaicite:29]{index=29}
  • The text quickly jumps from market observation to geopolitical thesis ("calculated market consolidation") without evidence for active steering. :contentReference[oaicite:30]{index=30}

7) Concrete additions that Clarus (in my view) should incorporate

  1. Clean up metrics

    • Instead of "5× hardware": Segment by server DRAM (RDIMM), HBM, SSD/NAND, complete servers; distinguish spot vs. contract. :contentReference[oaicite:31]{index=31}

  2. Add Swiss situation assessment

    • DC boom + power/acceptance question (SRF) as counterweight to "dying" thesis. :contentReference[oaicite:32]{index=32}

  3. Present Swisscom/Green differentiated

    • Swisscom as hybrid integrator (AWS/Azure) + CH colocation/"Swissness".
    • Green as hyperscaler-capable campus operator in expansion. :contentReference[oaicite:33]{index=33}

  4. Operationalize digital sovereignty

    • Jurisdiction (Cloud Act), exit strategies, key management, auditability. :contentReference[oaicite:34]{index=34}

Conclusion: My judgment in one sentence

The crisis/scarcity in storage and server components is real – but the Clarus article is too undifferentiated with the "5×" number and the "survival struggle of Europe"; for Switzerland, energy, location policy, and sovereignty governance are currently at least as decisive as DRAM prices.

Zurück zur Übersicht