Research & data studies
Original, cited reference reports on the infrastructure UK organisations actually buy — every figure sourced and dated. Free to read, free to cite.
The ICO cyber caseload: the published outcome of 956 closed cyber investigations, 2021 to 2026
The published trackers of ICO enforcement count the actions the Information Commissioner’s Office takes. None of them carries the cyber caseload those actions came out of, because the enforcement register does not contain it; the closest prior work, Bristows in April 2023, computed a caseload denominator for one calendar year on a mixed civil-and-cyber population and is credited throughout. We read all 41 of the ICO’s own quarterly “Cyber investigations” and “Cyber incidents” CSV files, January 2021 to March 2026, and computed the denominator: 956 closed cyber investigations, 100% outcome coverage, 956 distinct case references and no duplicate. 10 of the 956 (1.05%, one in 96) ended on the penalty track, 23 (2.41%) in a reprimand and 923 (96.55%) in neither, with “No action for DC” (518, 54.18%) and “Advice provided” (238, 24.90%) together just under four fifths of the caseload at 756 of 956 (79.08%). The pooled rate is published with its weakness beside it: 502 of the 956 closures (52.51%) fall in the single quarter January–March 2021, the first the ICO published, which contains no penalty-track case and no reprimand, so the same 10 cases are 2.2% of the 454 closures outside it and 8 cases are 4.76% of the 168 closed from January 2022 onwards. Why that quarter is so large is not something the published data answers and no explanation is asserted; the nearest evidence, published rather than left out, is that 287 of its 502 closures carry a case reference whose year is earlier than the closure year against 25 of 231 in the next quarter. Both ends are stress-tested: four defensible codings of the penalty track give 8 to 11 cases out of 956 (the published figure is 10, on Bristows’ grouping extended by one outcome string — on theirs unchanged it is 9), and removing the 81 administrative closures moves the rate from 1.05% to 1.14%. The escalation rate, any sector table, any duration metric and any pound figure are dropped, each with its reason on the page. 1,752-row dataset with the source file, URL and SHA-256 hash on every row, and no organisation named anywhere.
What CISA’s own ransomware advisories name: a census of the #StopRansomware series
CISA publishes 25 advisories under its own ‘#StopRansomware:’ banner. We read every one of them on 29 September 2026 and counted what is written in it. 11 of the 25 (44%) name no CVE identifier anywhere on the page as served, and reading each advisory’s current companion PDF as well returns the same 11 — the count only moves, to 10 of 25, if the superseded PDF revisions three advisories still link are read too. The 14 that do name a CVE give 64 mentions of 48 distinct identifiers, from one to eleven per advisory, and all 48 are in CISA’s own Known Exploited Vulnerabilities catalogue, all 48 flagged for known ransomware campaign use against a catalogue-wide rate of 361 of 1,728 (20.9%) — consistency between two CISA products, not independent corroboration, since CISA maintains both. Under Initial Access the series prints 62 advisory-technique pairs on a whole-document rule (59 on a stricter tables-only rule, the two differing on one advisory); T1190 Exploit Public-Facing Application leads at 17 of 25, and T1110 Brute Force is coded under Initial Access in none of them. 19 of the 25 publish a version history and 6 publish none at all, recorded as null rather than zero. CISA’s own curated StopRansomware index links 18 of the 25 and omits 7. Every figure is a count of what CISA’s documents name, never a statement about how ransomware works: an advisory naming no CVE is not evidence that the family exploits none. Population established three ways from CISA’s sitemap, its faceted advisory listing and its curated index, on each advisory’s own title — all three being CISA’s own indexes, so completeness is asserted against CISA’s published indexes rather than independently of them. 25-row dataset with a README, every row carrying its source URL, retrieval timestamp and page checksum.
The commercial gigabit gap: the UK premises that cannot order a gigabit broadband connection
Ofcom publishes fixed broadband coverage twice at every geography — once for all premises, once for residential premises only — and never publishes the difference below nation level. That difference is the commercial premise base. 600,791 of the UK’s 2,037,917 commercial premises (29.48%) cannot order a gigabit-capable broadband connection, against 11.08% of 30,455,777 homes: a gap of 18.40 percentage points, and a commercial premise 2.66 times as likely as a home to have no gigabit option. The national and per-nation figures are Ofcom’s own; what this study adds is the proof that the subtraction is exact (165 published cells in each release, none failing), the same figure derived for all 361 local and unitary authorities and all 650 Westminster constituencies, and the unmatched-premise bound. The commercial-specific penalty is proportionally sharpest in towns and cities: urban commercial premises are 3.33 times as likely as urban homes to be without, against 1.58 times rurally, and 68.90% of the whole shortfall is urban — though in percentage points the UK rural gap (19.38pp) is the wider one, rural commercial premises are still 19.38pp worse off than rural homes, and the page leads with both readings rather than the flattering one. These are UK figures including Northern Ireland, which runs the other way. Ofcom cannot match 1.551% of commercial premises to any provider record against 0.213% of homes, an asymmetry of 7.29 times computed from the counts, so the bound is published twice rather than caveated: assume every unmatched premise on both sides is gigabit-capable after all and the gap is still 17.07pp; move only the commercial side, which is the harder case for the finding, and it is still 16.85pp. Over six months the shortfall narrowed in percentage points (19.14pp to 18.40pp) and widened in relative terms (2.49 times to 2.66 times). A commercial premise is a property classification in Ordnance Survey AddressBase, not a count of trading organisations, and every column header, chart label and answer says so. No ranking appears without the published floor of 2,000 commercial premises. Ofcom data, Ofcom copyright, provided on an open basis via the Open Government Licence with no version named; 1,011-row dataset, 44 columns, no personal data and no premise-level record.
The UK public sector IT contract expiry cliff: 9,416 IT contracts reach their stated end date in 12 months
We harvested the whole published back catalogue of Contracts Finder (from 17 December 2014) and the Find a Tender Service (from 1 January 2021) from their daily bulk files and turned the axis round: not what was published, but what states an end date next. 9,416 UK public sector IT contracts state a contract end date in the 12 months to 24 September 2027 — out of 45,403 contracts of every kind ending in that window — held by 1,430 distinct buyers counted by normalised name, with 2,082 of them (22.1%) in March 2027 alone and 1,405 on 31 March 2027, the last day of the financial year. The 8,928 that state a positive sterling amount total £130.13 bn as published, and that figure is three things at once: a floor, because contracts stating zero, a negative amount, another currency or nothing are excluded rather than estimated and Find a Tender records a contract end date only on notices published from 24 February 2025; an over-count, because framework ceilings are counted alongside the call-offs placed under them; and not a measure of IT spend, because 52.1% of it sits on 270 contracts whose primary CPV code is not an IT division and the single largest notice in it is a £51.00 bn Crown Commercial Service energy supply framework. On the pre-registered primary-CPV sensitivity the same window is £62.09 bn. Lead on the count: holding the CPV basket still and changing only the field rule moves it by 3.5% while the money halves — though widening or narrowing the basket moves the count a great deal, from 4,764 to 9,945. The IT definition is a pre-registered CPV code list, hashed before any result existed, published with narrow and wide sensitivities. There is no supplier column, no supplier ranking and no supplier total. Forward expiry views of these feeds already exist commercially — PSIP, Stotles, Tussell — and are cited; what is published here is the method, and a 9,416-row dataset with a README.
Who holds a current NHS data-security assessment? A census of the DSP Toolkit register
Every one of the 93,260 registered entries on the NHS Data Security and Protection Toolkit’s public register on 25 September 2026, counted by whether an assessment has been published for the 2025-26 (version 8) cycle or later. A headquarters assessment can cover many listed sites, so an entry is not always a separate organisation. 29,791 (31.94%) have none; restricted to the 79,678 entries whose code is still Active in NHS ODS it is 17,477 (21.93%), and 5,436 (5.83%) have registered and never published anything at all. A missing assessment is not a finding of insecurity, and that is measured rather than asserted: 45.61% of the 24,132 lapsed entries that resolve in NHS ODS are marked Inactive there — 11,007 of them — against 1.55% of the 63,181 current entries that resolve. The study set out to count supplier compliance and reports instead, with the evidence, that the register cannot measure it — no 1,279-row supplier population exists, the label has been rewritten twice, and the only published definition of it, NHS England’s 2024-25 organisation-types guidance, sets a staff and turnover threshold against the retired label “IT Supplier” and sends everyone smaller to an “Other” bucket of 12,041 shared with charities. Cut by pooled sector, with both the pooled and unpooled readings published, and by Integrated Care Board: 13.17% (n = 2,210) to 27.09% (n = 4,795) across the 36 boards recorded on the register, against 31.12% of the 11,766 ODS-Active entries with no board recorded (63.04% of all 25,341 such entries, most of the gap being closure rather than geography). Registered is not required and every figure says so. Source register is Crown copyright under a free-reproduction permission, not the Open Government Licence; 93,260-row dataset, 21 columns, no names of registered organisations and no postcodes — the only names in it are the Integrated Care Board labels NHS ODS puts on a row — and the organisation code is not anonymous.
2.6% of the sites, 12.3% of the energy: data centres in the UK’s Climate Change Agreements scheme
The Environment Agency publishes who is certified for the Climate Change Levy discount, and separately publishes how much energy each sector reported. It does not join the two. Joined: data centres are 214 of the 8,285 facilities certified for certification period 7 (2.58%) and 12,327.58 of the 100,323.47 GWh the scheme reported for target period 6, the calendar year 2024 (12.29%) — an energy share 4.76 times the facility share, and 6.35 times the combined energy of the scheme’s 1,796 craft bakeries and 2,032 supermarket food-processing facilities. On energy per certified facility the sector ranks only 9th of 45. 214 counts members of one voluntary tax-discount scheme, not data centres in the UK, and certification is not a pass mark. Environment Agency data, Open Government Licence v3.0; 8,285-row dataset, 24 columns, no addresses.
One of 325 local authorities has an average half-hourly meter as big as it was in 2013: Slough
Of the 325 Great Britain local authorities with more than 100 half-hourly meters in 2013 and still present in 2024, exactly one has an average large-site meter drawing as much electricity now as it did then: Slough, 1.8892 GWh against 1.8878 GWh — parity by 0.08%, a margin thin enough that the revisions caveat matters. The median authority is at 29%, and much of that fall is settlement reclassification rather than falling demand: Great Britain’s own mean fell to a ratio of 0.301 over the same period. Over 2019 to 2023 — a window between the two documented settlement breaks, though half-hourly meters still rose 8.5% across it — 6 of 349 authorities are at or above their own 2019 average, 326 of 350 used less large-site electricity and 24 used more, and the ten largest risers took 94.6% of the 1,093.9 GWh of increases, with falls not netted off; the same ten rise on the migration-immune all-non-domestic series, where the top ten take 99.1% of 941.7 GWh over only 13 rising authorities. Recomputed against a large-site denominator rather than DESNZ’s all-electricity one that includes homes, data centres are 3.67% of Great Britain’s large-site electricity and 74.6% of Slough’s — an approximation, an all-meter numerator over a half-hourly denominator. All 2024 figures sit on a half-hourly meter base that jumped 23.4% that year, for reasons that are not explained in the DESNZ documents read for this study. DESNZ data, Open Government Licence v3.0; 378-row dataset, 47 columns.
Where UK government cloud data is declared to go: a census of G-Cloud 15
We read the self-declared data-location fields of all 14,468 services on G-Cloud 15’s three in-scope lots on 20 September 2026: 27.6% give the buyer no say in where data is stored or processed, 61.4% can hold or process it outside the UK and 38.0% beyond the UK and the EEA, and 49.2% of the in-scope catalogue is resold. Declarations, not locations. Full dataset, one row per listed service.
Windows 10, one year on: the security fixes a non-ESU PC did not receive
A Windows 10 Home or Pro 22H2 PC with no Extended Security Updates has not received 1,486 Microsoft security fixes since support ended on 14 October 2025 — 12 of them for flaws CISA lists as actively exploited, 9 listed on the day the fix shipped. Counted from the CVE Program’s own records; Microsoft’s Security Update Guide puts it at 1,500 for the same months. Enrolment stays open to 12 October 2027, so this is what a machine has not received to date. Full dataset and published filter.
FTSE 350 email security 2026: DMARC, SPF and MTA-STS checked against NCSC guidance
We checked the public email-authentication records of FTSE 350 companies on 19 September 2026 against NCSC guidance, using the email domain each company or trust publishes. Look up each named company and trust, with the NCSC fix for each gap.
UK public sector and VMware after Broadcom: a contract-notice tracker
UK public-sector contract notices naming VMware, Nutanix, Hyper-V or Horizon, found by keyword scans of Contracts Finder, Find a Tender, Public Contracts Scotland and Sell2Wales, 2021 to 2026: 620 procurements, with same-buyer renewal comparisons before and after Broadcom’s acquisition. Notice links and open data.
UK laptop and tech prices 2010–2026: what ONS consumer price data show
ONS’s quality-adjusted laptop index fell 73.9% from 2010 to 2026 (84% after inflation). In the 12 months to August 2026 desktop and tablet indices rose, and SD card and USB stick indices rose sharply. Chained series, price distributions and regional splits.
Windows Server 2016 and Windows 10 end of support: what UK public bodies’ own FOI answers show
Windows Server 2016 extended support ends on 12 January 2027. In 2024-25 Freedom of Information answers, 51 of 70 UK public bodies that named their Windows Server releases listed it. Aggregates only, from answers to requests others made.
UK data breach statistics 2026: seven years of reports to the ICO, by type and sector
We analysed all 77,222 personal data breach reports in the ICO’s published data, 2019 to 2025: 2025 was the highest year yet, most reports are not cyber attacks, and ransomware reports peaked in 2023. Sector tables with n, and a tidy dataset.
Oil prices, the 2026 conflict involving Iran and the cost of IT: what moved and what did not
Brent rose 27.8% from February to August 2026. We tested 93 official UK, US and EU price series: IT import and consumer prices did not follow oil, while fuel, plastics, aluminium and some freight did. HMRC data show 48–68% of UK IT import value arrives by air.
Which servers support both Windows Server 2025 and VMware ESXi 9? The 2026 census
We checked 334 Dell PowerEdge, HPE ProLiant and Lenovo ThinkSystem models against Broadcom, Microsoft and each maker’s own support matrix: 90% of current models are certified for both, 16% of 2017–20 models. Conflicts between sources reported, full dataset.
UK business e-waste statistics 2026: how much business IT does the official system see come back?
Environment Agency WEEE data 2008–2025 for business-only IT: 49,486 tonnes placed on the UK market in 2025, 802 tonnes recorded collected. Every figure traced to the source spreadsheets.
Which business laptops can you still upgrade? RAM, SSD and LPCAMM2 census 2026
We checked the manufacturers’ own service manuals for 412 Dell, HP, Lenovo, Apple and Microsoft business laptop variants (2023–2026): who still lets you add memory, swap the SSD or replace the battery. Every model sourced.
UK public sector email security census 2026: DMARC, SPF and MTA-STS against the GOV.UK standard
We read the email-authentication DNS records of 907 UK councils, NHS bodies, universities, police forces and fire services on 19 September 2026 and compared them with the government’s own email security guidance. Every organisation can look up its result, with a suggested fix for each gap. Dataset included.
Server Power Density 2026: the rack runs out of power, not space
Provisioned power per rack unit across 108 current Dell, HPE and Lenovo servers — how many fit a 6–60 kW rack, where accelerators change the arithmetic, and where liquid cooling starts. Dataset included.
Dell vs HPE vs Lenovo: how much of a server is actually the vendor?
Every processor and accelerator mapped across the three catalogues: 68% of processors are sold by all three. What you are really choosing is the chassis, the ceiling and the management stack. Dataset included.
AI Servers: The 2026 Data Study
A cited, UK-focused reference on AI/GPU servers — architecture, GPU specs (H100/H200/B200/MI300X), the vendor landscape, power & cooling, cloud-vs-own economics and the market backdrop.
UK Leased Line Price Guide 2026: Real Costs by Region
Real 2026 UK leased line prices for 100Mb, 1Gb and 10Gb across London, cities and rural areas, plus hidden costs and a 36-month TCO worked example.
Server Carbon Footprint 2026: The Vendor PCF Dataset
September 2026 server carbon footprint dataset compiling Dell, HPE, and Lenovo PCF disclosures, operational splits, and UK procurement guidelines.
Server Failure Rates 2026: Component-Level Field Data Study
2026 server failure rates analysed: Backblaze Q1 fleet AFR (1.24%) vs Google DRAM error rates (>8%), paired with UK channel drive pricing and spares models.
Refurbished IT Market Size 2026: What the Data Does (and Doesn't) Show
2026 refurbished IT market analysis: evaluating the £7.6bn UK trade estimate against component price surges, DRAM inflation, and doubling client-PC volumes.
Server Warranty Length & Terms Compared: 2026 OEM Dataset
Compare standard server warranty length, parts versus labour terms, and response tiers across major OEMs and UK secondary market specialists in 2026.
DDoS Attack Statistics 2026: Mid-Year Analysis
Mid-2026 DDoS attack statistics synthesize Cloudflare and Akamai telemetry with transit insights, attack vectors, EMEA volumes, and UK compliance guidance.
Phishing Statistics UK 2026: Official Breach & Threat Data
Official UK phishing statistics from DSIT and NCSC: 2025/26 breach rates, weekly attack frequencies, sole-vector figures, and national reporting volumes.
Microsoft Vulnerabilities Per Year: 2026 Stats and Historical Trends
This analysis of Microsoft vulnerabilities per year examines CVE trends (2020–2025), 2026 volume surges, and UK Cyber Essentials 14-day compliance.
UK Cyber Security Industry Statistics 2026: Revenue & Jobs
DSIT 2026 UK cyber security industry statistics show around £14.7bn revenue, 2,603 firms, approximately 69,600 FTE jobs, and regional employment analysis.
Cyber Security and Resilience Bill Timeline: 2026 Tracker
Mid-2026 tracking data on the UK Cyber Security and Resilience Bill timeline, statutory 24-hour and 72-hour reporting clocks, and phased commencement.
Ransomware Backup Targeting Statistics 2026: The Data Study
Analysis of 2026 backup targeting and compromise statistics: how adversaries attack repositories, actual recovery rates, and UK regulatory compliance.
How Long Does Data Restore Take? Restore Speed Index 2026
Compare data restore speeds across LTO-9/10, disk, and cloud. Access benchmarks, throughput data, and UK compliance guidance to verify your true RTO.
Microsoft 365 Retention Limits and Shared Responsibility 2026
Microsoft sets Exchange and SharePoint deletion recovery at 14–93 days. Review native retention limits, shared responsibility boundaries, and UK GDPR rules.
Backup Software Market Share 2026: Revenue, Scale and Vendors
Mid-2026 backup software market share data: vendor revenues, Cohesity-Veritas scale, Rubrik's $1.32bn year, Veeam's lead, and UK cyber resilience rules.
UK Power Cost & Continuity Planning 2026
Mid-2026 UK power cut and grid resilience analysis: regional standing charges, GB unit rates, and data-backed UPS runtime planning for business continuity.
Major Cloud Outages of 2026: Incident Tracker & UK Impact
A tracker of major cloud outages 2026 across Azure and Microsoft 365, plus AWS workload case studies, with verified incident durations and UK impact.
Data Centre Outage Statistics 2026: UK & Global Trends
Data centre outage statistics for 2026: Uptime Institute findings, power failure root causes, UK grid constraints and the financial impact of downtime.
Refurbished Storage Array Prices: 2026 Price Index
Mid-2026 price index for refurbished storage arrays and enterprise drives, tracking NAND contract surges, secondary-market dynamics, and UK TCO.
SSD vs HDD Power Consumption 2026: UK Watts-to-£ Formula
Verified 2026 UK electricity rates turned into a watts-to-pounds formula for SSD and HDD storage — and why no cross-vendor watts-per-TB dataset yet exists.
S3 Storage Cost Per TB: 2026 Object Storage Index
Benchmark S3 storage cost per TB in 2026. Compare full workload pricing across AWS, Azure, and alternatives including request, retrieval, and egress fees.
NAND Flash Price Tracker 2026
Track 2026 NAND flash contract prices and UK enterprise SSD procurement benchmarks from TrendForce forecasts and verified price indices.
Hard Drive Failure Rates 2026: Backblaze Stats & UK Guide
Mid-2026 data study on hard drive failure rates. Analysis of 340,000+ Backblaze drives, 20TB+ reliability, SSD failure mechanisms, and UK storage pricing.
How Much Data Is Created Every Day? (2026 Figures, Corrected)
Corrected 2026 data growth stats: an estimate derived from IDC's Global DataSphere figures, plus UK commercial electricity and colocation costs.
LTO Tape Roadmap 2026: Shipments, LTO-10 & UK Pricing
LTO shipment data, LTO-10's 30TB/40TB capacities, the roadmap to LTO-14, and UK tape pricing from about £28 to just over £319 — the 2026 market tracker.
ARM Server Market Share 2026: Why There's No Single Number
ARM server market share in mid-2026: why no single percentage exists, what Graviton, Grace and Ampere data show, and UK on-prem procurement realities.
How to Find UK Government IT Hardware Contract Data (Sept 2026)
Government IT spending UK data for 2026: central department budgets, CCS frameworks and how to find individual server and infrastructure contract notices.
VMware Market Share 2026: Migration Data and Scoreboard
VMware market share in 2026: installed base metrics, Broadcom sentiment shifts, migration targets and real displacement rates across enterprise estates.
Core Infrastructure Software End of Life Calendar 2026–2036
An enterprise software end of life calendar for Windows Server, SQL Server, RHEL, Ubuntu and vSphere from 2026 to 2036, with UK compliance guidance.
CPU Price Per Core Index 2026: Xeon vs EPYC
Mid-2026 server CPU price per core benchmark: Intel Xeon vs AMD EPYC 1k-unit list prices, exposing cost sweet spots, July 2026 repricing, and UK TCO.
Q1 2026 Server Market Data: Statistics, AI Splits & UK ASPs
Server market size 2026 from Q1 data: revenue hit $122.6bn, non-x86 reached 47.9%, and UK buyers face inferred ASP pressure and memory supply constraints.
Server Power Consumption: 2026 SPECpower Efficiency Index
Discover 2026 server power consumption benchmarks. We analyse official SPECpower results, compute performance per watt, and model UK electricity costs.
HBM Market Share Tracker 2026: Supply, Share and Pricing
Mid-2026 HBM market share tracker analysing SK hynix, Samsung, and Micron capacity, HBM4 qualification timelines, and UK AI infrastructure pricing.
Server CPU Core Count Trend: 2026 Density and Licensing
Analysis of the server CPU core count trend in 2026, tracking density jumps up to 192 cores per socket and the resulting UK enterprise licensing squeeze.
AI Server Market Share Tracker 2026: Dell vs Supermicro
Q1 2026 server revenue data: ODM share fell to 50.2%, Dell hit 16.5% ($20.28bn). Full vendor market-share tracker built for UK AI infrastructure buyers.
GPU Residual Values 2026: The UK Depreciation Study
Verified 2026 data tracks used H100, H200 and A100 prices against launch cost — real depreciation curves and landed-cost guidance for UK IT buyers.
LLM API Price Tracker 2026: Cost Per Million Tokens
Dated August 2026 LLM API pricing comparison: verified cost-per-million-token figures for GPT, Claude and Gemini tiers, with UK buyer analysis.
AI Accelerator Comparison 2026: Nvidia vs AMD vs Intel
Normalised Nvidia B200, DGX B200 and Intel Gaudi 3 specs from vendor datasheets — HBM, bandwidth, TDP, FP8/FP4, fully footnoted for UK AI buyers in 2026.
UK AI Data Centre Build-Out Tracker 2026: Five Zones Mapped
A sourced tracker of UK AI data centres in 2026: AI Growth Zone announcements by site, pledged investment, MW and planning status, checked against sources.
UK AI Adoption Statistics 2026: The Verified Numbers
ONS, DSIT and Bank of England data reconciled: the real 2026 UK AI adoption rate by company size, plus depth-of-use, barriers and regulatory context.
UK SME Cyber Attacks 2026: The Real Frequency & Cost
UK SME cyber attack data for 2026: 43% of UK firms breached, phishing dominant, £6,400 average SME breach cost. Segment-by-segment figures inside.
RTO/RPO Reality Index 2026: Stated vs Actually Achieved
UK RTO/RPO benchmark 2026: real DR drill data shows recovery in 73–121 seconds vs vague policy targets. See the gap — and how to close it.
Lease vs Buy Servers 2026: The CapEx-OpEx Maths Reworked
UK data study: lease vs buy servers 2026. Real TCO figures, utilisation thresholds and UK tax rules for the CapEx-vs-OpEx call in a price surge.
ITAD Value Recovery 2026: What Retired IT Is Really Worth
ITAD value recovery 2026 data study: what retired IT is worth by asset age, resale vs sanitisation cost, and the WEEE duties UK teams need to plan for.
Phishing-Resistant MFA 2026: Passkeys vs OTP Data Study
UK data study: passkey and FIDO2 adoption figures for 2026, Microsoft's Entra ID deadline, and why OTP/push MFA is being retired.
Backup Window 2026: The Data Growth vs Throughput Gap
2026 data study: enterprise data growth vs backup throughput. UK figures on the shrinking backup window and how to close it.
AI Networking Index 2026: 800G Goes Mainstream
800G data centre networking 2026 tracked with verified adoption, cost-per-port and Ethernet vs InfiniBand data for UK buyers planning AI cluster upgrades.
Cold Storage £/TB 2026: Tape vs HDD vs Object Cost
Mid-2026 £/TB data for tape, HDD and cloud archive tiers — full TCO, hidden fees and UK G-Cloud pricing for 10-year retention decisions.
CVE Exploitation Speed Index 2026: The Patch-Gap Race
2026 data study: mean time to exploit a CVE has hit -7 days. UK-focused figures on the patch gap, AI impact and how to prioritise remediation.
IT Hardware Financing 2026: Monthly Cost of the Price Surge
UK data study: 2026 hardware price rises of 15-30% meet elevated financing costs. Worked monthly-cost examples for servers, laptops and leasing.
Server End of Service Life 2026: The UK EOSL Dataset
UK data study: exact Windows Server EOSL dates for 2026-27, what mainstream vs extended support really means, and how to sequence your server refresh.
OEM Support Price-Hike Tracker 2026: UK Cost Data
2026 OEM support renewal price hikes tracked by vendor, with UK figures, a worked TPM comparison, hidden costs and a renew-or-switch framework.
AI Inference Energy Cost 2026: Pounds per Million Tokens
Mid-2026 UK data study: GPU wattage, UK electricity benchmarks and real £/$ per-million-token pricing — a working formula for AI inference cost.
DRaaS Price 2026 UK: Benchmarking Cost, RTO and Failover
UK DRaaS pricing benchmarked for 2026: per-VM rates, sovereign list prices and RTO/RPO tiers compared, so buyers match spend to workload risk.
Largest SSD 2026: Enterprise Capacity Index Past 122TB
Largest SSD 2026 index: 122TB-class drives, bigger models shipping or announced, EB-scale shipment growth and the UK pricing reality for enterprise buyers.
LLM VRAM Requirements 2026: The Open-Model GPU Index
Exact 2026 VRAM and GPU-tier figures for Llama 4, DeepSeek V4, Kimi K2.5 and Qwen 3 32B — weights, KV cache and quantisation tiers for UK IT buyers.
Network Switch Prices 2026: The UK Procurement Tracker
Verified 2026 data on Cisco price increases, switch lead times and firewall costs — plus a UK procurement playbook for network refreshes.
UK Cyber Insurance Premium Index 2026: Costs & Controls
2026 UK cyber insurance data: premium bands by turnover, broker rate trends, required controls, and the £25,000 Cyber Essentials cover bundle explained.
All-Flash vs Hybrid TCO 2026: The 60–80% Power Argument
Seagate puts SSD TCO at about 6x HDD; MarketIntelo says all-flash cuts power per TB 60–80% in dense AI. Where flash wins and where hybrid still rules.
Data Centre Liquid Cooling 2026: The kW Where Air Dies
UK data centre liquid cooling 2026: the exact kW thresholds where air fails, direct-to-chip vs immersion bands, and what GB200-class racks mean for retrofits.
Immutable Backup Adoption 2026: UK Data Study
2026 data study: UK immutable & air-gapped backup adoption rates, the insurance-driven mandate, and the recovery-time gap IT buyers must close.
End of Support Attack Surface 2026: The Unpatched Wave
Windows 10's consumer security bridge closes 13 Oct 2026. A data-led 2026 risk study for UK IT leaders on ESU deadlines, compliance and cost decisions.
Xeon 6 vs EPYC Turin 2026: Cores, Watts & UK Pricing
Xeon 6 vs EPYC Turin 2026 compared with vendor-published core counts, AMD/Intel benchmark claims, US list prices and TCO factors for server buyers.
Cloud Repatriation 2026: The UK Sovereign AI Data Study
UK data study: 87% of firms plan cloud repatriation, 93% move AI data UK-only, and the TCO math behind sovereign AI and hybrid infrastructure decisions.
Third-Party Maintenance Savings 2026: A UK Data Audit
Our 2026 audit of third-party maintenance savings claims for UK IT buyers: what's verified, what isn't, and how to build a defensible TCO case instead.
Nvidia Rubin GPU Specs 2026: The UK Buyer's Roadmap Tracker
Nvidia's Rubin roadmap decoded: specs vs Blackwell Ultra, power/cooling impact for UK data centres, and when procurement teams should actually buy.
Largest Hard Drive 2026: HDD Roadmap Tracker (UK)
WD's 40TB HDD nears volume production as Seagate ships 44TB-class drives to hyperscalers. Full 2026 HAMR/ePMR/MAMR roadmap for UK buyers.
Backup £/TB Tracker 2026: The Storage-Crisis Knock-On
UK backup £/TB data study: G-Cloud rates, hyperscaler tiers, and specialist pricing compared for 2026, with worked TCO examples and a buyer checklist.
Server Lead Times 2026: The UK Delivery-Time Tracker
Server lead times in 2026: published figures put general x86 servers at 30–52 weeks, AI GPU systems 40–70+ weeks and DRAM 40+ weeks, plus Q1 2026 price rises.
UK Data Breach Cost 2026: The Real Pound Figure
UK data breach cost 2026: £3.29m average, £5.74m in financial services, and why DSIT's £0 median masks the real board-level exposure.
On-Prem AI Inference Break-Even 2026: Own vs Rent UK
UK data study: on-prem AI inference costs £75k–£90k/yr vs £200k–£240k cloud for an 8-GPU A100 node. Where the break-even utilisation line falls in 2026.
Server Refresh 2026 TCO: Refresh, Defer or Extend?
UK data study: 2026 server refresh TCO modelled against deferral and TPM extension, using IDC, Servnet and circular-economy refresh research.
GPU Rack Power Density 2026: 600W Chips to 600kW Racks
2026 data: GPU TDPs to 1400W, AI racks at 80-120kW, GB200 NVL72 at 120kW. The UK buyer's power-density chart, worked example and retrofit costs.
Windows Server 2016 EOL: The January 2027 Migration Dataset
Windows Server 2016 support ends 12 Jan 2027. UK data on ESU rules, installed base and migration timelines for IT teams planning now.
2026 Nearline HDD Shortage: AI Reprices UK Cold Storage
The 2026 nearline HDD shortage for UK businesses: how AI demand is driving storage prices and lead times, and practical strategies for procurement.
Refurbished Server Price Index 2026: The Secondary Surge
Refurbished vs new servers in 2026: sourced figures on DRAM/NAND contract prices, new-server price rises, CPU supply and used-server pricing, plus a buyer checklist.
2026 Storage Price Index: What the Crisis Did to Arrays
UK data study: NAND, SSD and HDD price moves in 2026, tracked against enterprise storage spend. See the numbers behind the storage cost shock and what to do next.
SSD vs HDD Cost Per TB 2026: The 22.6x Capacity Gap
Verified Q1–Q2 2026 data: 30TB enterprise QLC SSDs cost 22.6x as much per TB as HDDs. UK-specific TCO, power and market analysis for storage buyers.
UK GPU Rental Price Index 2026: H100 to B200 Hourly
Verified mid-2026 data: H100, H200 and B200 rental rates from $0.36 to $14.90/hr. See the true cost gap and what it means for UK AI buyers.
Windows 10 EOL 2026: The Billion-PC Refresh, By The Numbers
Windows 10 EOL 2026 data study: ESU deadlines to 2027/2028, real commercial pricing, Windows 11's billion-device flip, and the UK refurb-PC surge.
DDR5 Server Memory Price 2026: The 90% Quarter Tracked
DDR5 server memory price 2026 tracked by quarter: Q1's 93-98% DRAM contract spike, UK kit prices up 344%, and what Q3-Q4 forecasts mean for budgets.
Ransomware Recovery Cost Index 2026: Downtime in Pounds
Ransomware recovery cost 2026 in pounds: what UK mid-market firms pay per incident by size, typical downtime days, and how immutable backups cut the bill.
Server Prices 2026: How Memory Repriced the Rack
Tracking 2-socket server prices through the 2026 memory shock: DRAM up 90-95% in Q1, Dell list +17%, and a UK-indexed 512GB reference build.
2026 AI Server Cost Index: The HBM Tax on GPU Boxes
AI server cost 2026, decoded: a GBP teardown of an 8xH200 GPU box shows the HBM and DRAM memory tax now driving most of the bill. Cited figures.
Research is one of the four article libraries in the Servnet Hub. The others are Insights (long-form technical guides), Learn (plain-English explainers) and News (same-day analysis).
Talk to a UK specialist
Get expert advice or a no-obligation quote — servers, storage, networking, maintenance, finance and cloud. We reply the same working day.