Everpure FlashBlade: Object Storage Ledger 2026 Scorecard

Part of the Object Storage Ledger (edition 2026-10-16). How the scores work: methodology.

Snapshot

Type: product
Owner: Everpure, Inc. (NYSE: PSTG, USA), the new name of Pure Storage, Inc.: certificate of amendment filed 2026-02-20 (SEC 8-K), announced 2026-02-23 with an intent to acquire 1touch, trading as Everpure from 2026-03-05 with the ticker unchanged; the rename FAQ says leadership, contracts, SKUs, support and product functionality do not change
Licence: Commercial; FlashBlade hardware with Purity//FB under Evergreen subscriptions, or Evergreen//One consumption (pay for use, now extended to FlashBlade//EXA); prices not public
Latest release: Purity//FB 4.6.8 (current on 2026-03-17); 4.7.10, the first long-life release of the 4.6/4.7 line (GA date not found)
Support lifecycle: Feature releases and long-life releases; the 4.6/4.7 line is planned to reach end of life in April 2029 (community post via search summary); Evergreen promises non-disruptive upgrades of blades and software (vendor)
Next milestone: none with a date; Everpure Data Stream in beta sign-up (2026-03-17)
Object lock: Purity//FB 4.1 (blog 2023-02-12): retention per object or bucket, legal hold, bucket default retention (1 to 365,000 days in the Ansible module), a ratcheted retention lock that only Pure Technical Services can lower and an optional freeze of locked objects; works on versioned and unversioned buckets; mode names not printed in sources read; Object Lock on the third site of fan-in replication from 4.1.5 (2023-08-08). Veeam Ready Capacity Target with Immutability for //S200 on 4.1.0 (sys504), //E (sys698) and //S100 on 4.5.6 (sys1510), each noting IAM and STS as not supported; a Commvault best practices paper (search result); no independent compliance assessment found
Tier: File-first platform with S3
Archetype: Dependable builder, hardware-bound
Evidence grade: B. The SEC 8-K and the Everpure rename FAQ, three Veeam Ready listings, the Purity//FB 4.1 and 4.1.5 blogs, the 2025-04-04 S3 over RDMA blog, the March 2026 launch post and the Ansible collection docs were read directly. The 4.6 and 4.7.10 release details come from search summaries of community posts that returned HTTP 403 (grade B). The first release year, the erasure coding scheme and all steps before 2023 are grade C.

Scope of the scores: FlashBlade is sold only as Everpure hardware running Purity//FB, and it is file-first: NFS, SMB and S3 share one operating system. The scores cover the S3 object side; the hardware coupling shows in deployment freedom and lock-in. The company was Pure Storage until the rename filed on 2026-02-20, and the rename FAQ says the product and its contracts did not change. The first scored year, 2017, is background knowledge (grade C).

Scorecard

Dimension	Score	Direction
Capability today	63.0 / 100	higher is better
Momentum since end of 2022	+16.4 points (2017 to 2022: +6.1)	higher is better
Credibility	16.0 / 25	higher is better
Commercial risk	12.0 / 25	lower is better
Technical lock-in	14.5 / 25	lower is better
Integrator fit	14.0 / 20	higher is better
AI leverage	7.0 / 20	higher is better
Ledger Index	58.9 / 100	balanced weights; see the methodology

Capability by domain, 2015 to today

Scores 0 to 5 against fixed anchors; total is weighted to 100.

Domain (weight)	2015	2019	2022	Today
S3 API fidelity (15)	n/a	3	3	3
Immutability and compliance (12)	n/a	1	1	3.5
Durability and resilience (12)	n/a	2.5	2.5	4
Performance and scale (12)	n/a	3.5	3.5	4.5
Deployment freedom (10)	n/a	1	1	1
Security and multi-tenancy (10)	n/a	2.5	2.5	3
Operations (10)	n/a	3	3	4
Protocols beyond S3 (7)	n/a	3.5	3.5	3.5
Lifecycle and tiering (7)	n/a	1	1	2
Ecosystem and openness (5)	n/a	1.5	1.5	1.5
Total / 100	n/a	46.6	46.6	63.0

Year by year: 2017: 40.5 · 2018: 40.5 · 2019: 46.6 · 2020: 46.6 · 2021: 46.6 · 2022: 46.6 · 2023: 58.8 · 2024: 58.8 · 2025: 59.8 · 2026: 63.0

What moved the score

Year	Domain	Change	Points	Trigger
2019	Immutability and compliance	0 → 1	+2.4	Veeam Ready Capacity Target (2019-05-15) without immutability
2019	Protocols beyond S3	3 → 3.5	+0.7	SMB beside NFS and S3 on one operating system (year approximate, grade C); 4.6.8 adds local users and groups for mixed SMB and NFS access; HDFS, Swift and Azure Blob not found
2019	S3 API fidelity	2 → 3	+3.0	Veeam Ready Capacity Target listing (2019-05-15), which needs working multipart and standard S3 semantics; versioning in the Ansible bucket module; IAM and STS not supported per Veeam; lifecycle, SSE and notifications not verified
2023	Durability and resilience	2.5 → 4	+3.6	4.1.5 (2023-08-08): active-active object replication between two sites, fan-out, three-site fan-in and replication to AWS or other S3 targets; failure domains and rebuild tests not read
2023	Immutability and compliance	1 → 3.5	+6.0	4.1: Object Lock with legal hold, default retention and a ratcheted retention lock; Veeam Ready with Immutability on //S200 with 4.1.0; 4.1.5 keeps the lock on the third replication site; mode names not printed; no independent assessment
2023	Lifecycle and tiering	1 → 2	+1.4	4.1.5: replication to AWS or other S3 targets, which copies rather than tiers; S3 lifecycle rules and quotas not verified
2023	Performance and scale	3.5 → 4	+1.2	FlashBlade//S and //E: Veeam test of 2 PB with a 4 PB maximum on //E (2023-09-26); 55 GB/s per chassis and up to 650 GB/s on //S (vendor)
2025	Security and multi-tenancy	2.5 → 3	+1.0	4.6 line: object realms for secure multi-tenancy, management access policies and object auditing (search summary, year approximate); 4.6.8 adds replication parity for realms; IAM and STS not supported per Veeam; KMIP and FIPS not found
2026	Operations	3 → 4	+2.0	4.6.8: proactive Purity patching; 4.7.10 long-life release with end of life planned April 2029; purestorage.flashblade Ansible collection; Terraform and fleet tooling not verified
2026	Performance and scale	4 → 4.5	+1.2	4.7.10 long-life release lists S3 over RDMA (preliminary 250 GB/s on five chassis, vendor, 2025) and FlashBlade//EXA (10+ TB/s, vendor); GA date not found; MLPerf, SPECstorage and IO500 results cited but not read

Direction, 2023 to 2026

Each release item carries one theme and one driver (V vision, C customer need, M market window, U upstream, P portfolio).

Date	Item	Theme	Driver
2023-02	Purity//FB 4.1: S3 Object Lock with legal hold, default retention and a ratcheted retention lock	Immutability	C
2023-05	Veeam Ready Capacity Target with Immutability for FlashBlade//S200 on 4.1.0	Immutability	M
2023-08	4.1.5: active-active object replication, three-site fan-in, Object Lock on the third site	Durability	C
2023-09	Veeam Ready listing for FlashBlade//E (2 PB tested, 4 PB maximum)	Ecosystem	M
2025-04	S3 over RDMA announced for later in 2025, preliminary 250 GB/s on five chassis	Performance	V
2025-05	Veeam Ready listing for FlashBlade//S100 on Purity//FB 4.5.6	Ecosystem	M
2026-02	Pure Storage renamed Everpure; intent to acquire 1touch; move beyond storage to data management	Packaging	P
2026-03	4.6.8: proactive Purity patching, local users and groups for SMB and NFS, replication parity for multi-tenant realms	Operations	C
2026-03	Everpure Data Stream beta for AI data pipelines to GPU clusters	AI	V
2026-10	Long-term outlook names Scale AI with FlashBlade//EXA as one of four growth vectors	AI	V

Centre of gravity: Immutability (2), Ecosystem (2), AI (2)
Blind spots (no items): S3 API, Deployment, Security, Protocols, Lifecycle, Licensing
Driver mix: Vision 30%, Customer need 30%, Market window 30%, Upstream 0%, Portfolio 10%
Idea-to-need index: 0.33 (0 = demand-pulled, 1 = thesis-pushed)

Credibility: 16.0 / 25

Criterion	Score	Basis
Cadence	3.5	Feature releases and long-life releases with GA posts (4.5.6, 4.5.7, 4.6.6, 4.6.8, 4.7.10); the GA dates themselves could not be read
Roadmap transparency	2	Blogs and launch posts public; community GA posts and the support portal behind a login; no roadmap
Say-do	3.5	S3 over RDMA, announced 2025-04-04 for later that year, is listed in the 4.7.10 long-life release; its GA date was not found
Velocity	3.5	Object realms, S3 over RDMA, //EXA support and proactive patching within the 4.6 and 4.7 lines
Lifecycle stability	3.5	Long-life releases with a planned end of life in April 2029 and Evergreen non-disruptive upgrades; the company rename moves product names through 2026

Commercial risk: 12.0 / 25 (lower is better)

Criterion	Score	Basis
Price and licence volatility	2	Evergreen model unchanged in sources read; the rename FAQ says contracts and SKUs do not change; prices not public
Purchase constraints	3.5	Hardware purchase with an Evergreen subscription or Evergreen//One consumption; no software-only edition
Channel and access	3	Partner led through Everpure Partner Central; support portal and community posts behind a login
Owner stability	1.5	Large listed company; renamed Everpure in 2026 with a move beyond storage and an announced acquisition of 1touch; leadership unchanged
Cost of staying supported	2	Evergreen promises non-disruptive upgrades of blades and software; the long-life line runs to April 2029

Technical lock-in: 14.5 / 25 (lower is better)

Criterion	Score	Basis
Formats	3	Proprietary Purity//FB layout; data leaves through NFS, SMB or S3
Export path	1	NFS, SMB and S3 reads plus replication to AWS or other S3 targets (4.1.5)
Stack coupling	3.5	Own operating system on own hardware; nothing third-party underneath
Hardware / cloud coupling	4.5	FlashBlade hardware only; no VM, cloud or software-only form; no mandatory cloud attach found
Skills and tooling coupling	2.5	Vendor tooling with an open Ansible collection; a sizeable partner pool

Integrator fit: 14.0 / 20

Criterion	Score	Basis
Partner programme openness	3.5	Channel-led Everpure Partner Central; terms not read
Multi-tenancy and self-service	3.5	Object realms for secure multi-tenancy with replication parity (4.6.8) and management access policies (search summary)
API and infrastructure-as-code	3.5	REST API, CLI and the purestorage.flashblade Ansible collection with bucket, lock and replication parameters; Terraform not verified
Skills and certification	3.5	FlashBlade certification exams exist (search result, grade B)

AI leverage: 7.0 / 20

How the product uses AI, scored per year like capability. Only GA counts.

Criterion	2019	2022	Today	Latest step
AI workload support	1.5	1.5	3.5	2026: 4.7.10 long-life release lists S3 over RDMA; FlashBlade//EXA for AI factories and NVIDIA DGX SuperPOD named (vendor); GPUDirect not named
AI platform stack	0	0	0.5	2026: Everpure Data Stream in beta sign-up (2026-03-17) to feed AI data pipelines to GPU clusters; half a step for a beta
AI-assisted operations	1	1	1	2017: Vendor monitoring with analytics and patch advisories (grade C); ML forecasting not verified this pass
Agent openness	2	2	2	2017: REST API and CLI, later an Ansible collection; no MCP server found

Year by year: 2017: 4.5 · 2018: 4.5 · 2019: 4.5 · 2020: 4.5 · 2021: 4.5 · 2022: 4.5 · 2023: 4.5 · 2024: 4.5 · 2025: 4.5 · 2026: 7

AI say-do, 2023 to 2026: 2 AI items announced; 1 GA, 1 still preview, 0 pending, 0 dropped. S3 over RDMA counts as GA because the 4.7.10 long-life release lists it; Data Stream is a beta. The //EXA platform and the 2026-10 growth outlook are product and business news with no AI feature to check.

FlashBlade's AI case is throughput: S3 over RDMA reached the 4.7.10 long-life release about a year after it was announced, and //EXA targets large GPU clusters on vendor figures. There is no AI platform beyond a Data Stream beta, no AI assistant in operations that this pass could verify, and no MCP server.

Reading

FlashBlade scores on durability, performance and operations: active-active object replication since 4.1.5, all-flash throughput with S3 over RDMA in the 4.7.10 long-life release, and a release scheme with a published end of life in April 2029. Immutability is solid on the evidence read, with a ratcheted retention lock, default retention and three Veeam listings, though no source printed the lock mode names. Deployment freedom scores 1 because it runs only on Everpure hardware, and the same fact drives most of its lock-in. A buyer should ask Everpure to confirm that both governance and compliance modes exist and whether IAM and STS are still unsupported, since all three Veeam pages list them as not supported.

Open questions and items to re-score

First release year (2017 is background knowledge, grade C) and the release dates of 4.6.x and the 4.7.10 long-life release.
Governance and compliance mode names and how bucket default retention behaves across modes.
S3 over RDMA GA date and client requirements.
Erasure coding scheme, failure domains and rebuild behaviour.
S3 lifecycle rules, SSE with an external KMS, KMIP, FIPS and event notifications.
Completion of the 1touch acquisition and whether FlashBlade product names keep the Everpure prefix.
Vendor analytics and any ML forecasting in operations were not read.