Drivenets

Company sizes or stages where this product is overkill or insufficient

• Mid-market enterprises without massive routing scale or AI cluster sizes where standard switching/routing platforms are simpler and cheaper end-to-end.

Industries or use cases where competitors are clearly superior

• Environments where a mature, widely-deployed NOS and a deeply established support ecosystem is the overriding priority (often Cisco/Juniper wins here).
• AI deployments standardized on NVIDIA’s end-to-end AI stack where Spectrum-X’s tight coupling is desired over multi-vendor openness. (nvidia.com)

Technical requirements that this product cannot meet

• Not proven from public sources, but buyers should be cautious if they need very specific legacy features, rare MPLS/traffic-engineering behaviors, or compliance certifications not publicly documented.

Budget constraints that make alternatives more appropriate

• If your biggest cost is staff time/complexity rather than hardware, a traditional integrated router platform may be cheaper overall.
• If you lack automation expertise, you may pay heavily for professional services and training (DriveNets provides certification). (get.drivenets.com)

Specific workflows or integrations where the product falls short

• Not well evidenced publicly; but “integration with existing OSS/BSS, telemetry pipelines, and change management” is a common friction point for disaggregated platforms.

Red flags in the buying process

• Limited independent review depth available publicly (harder to validate day-2 operations outcomes).
• Vendor claims on performance/economics without independently published benchmarking methodology.

Current limitations acknowledged by the vendor

• No explicit “limitations” section was captured from DriveNets public pages in this session.

Features on the roadmap (not yet available) — note estimated dates if available

• Not captured.

Scale limitations (at what volume/size does it break down?)

• Vendor claims scale up to very high Tbps cluster sizes for routing and large-scale GPU fabrics; however, buyers should validate real constraints:
  • topology constraints
  • buffer behavior and tail latency for AI
  • control-plane scaling and convergence times
  • operational tooling limits at thousands of nodes/links

Security or compliance gaps

• The main gap is lack of publicly documented certifications and controls in accessible sources; this does not mean they don’t exist, only that they were not found publicly.

Technical debt or architectural limitations mentioned in reviews

• The single G2 review mentions “stability issues” and “connectivity interruptions” after implementation. (g2.com)

API limitations or integration constraints

• Not captured in public sources.

• The strongest validation is not consumer-style reviews but large operator deployments and repeated independent coverage noting adoption by Tier-1 providers (e.g., AT&T deployment). (drivenets.com)

• DriveNets positions its architecture as a distributed “router as a cluster” scaling to very high Tbps levels using OCP DDC and merchant silicon, with orchestration as a first-class component. (drivenets.com)
• For AI: “fabric scheduled Ethernet” and end-to-end stack optimizations (claims should be benchmark-validated in your environment). (drivenets.com)

• Insufficient review volume to make a review-grounded differentiation claim. The single G2 review supports “ease of use/modular architecture” but also flags stability, high pricing, and learning curve. (g2.com)

• Evidence: AT&T production deployment (vendor PR + independent reporting referencing AT&T). (drivenets.com)
• Vendor economics claims exist historically (e.g., “40% cost savings” mentioned in TechCrunch coverage), but require buyer-side validation. (techcrunch.com)

• Cisco/Juniper:
  • Broader publicly documented security/compliance packaging and mature enterprise procurement processes.
  • Deep feature maturity and extremely broad install base.
• NVIDIA (AI Ethernet):
  • Highly optimized, tightly integrated AI networking platform approach (may deliver predictable outcomes when you buy into the NVIDIA stack).

Review platform coverage and limits

• G2 shows only one DriveNets review in the captured page, which is an extremely small sample and not representative. (g2.com)
• Gartner Peer Insights: an attempt to locate DriveNets-specific Peer Insights reviews did not yield an accessible, clearly relevant DriveNets product page in the captured sources; the accessible Gartner pages seen were either general Peer Insights landing pages or unrelated “Driven Technologies” content (not DriveNets). Treat Gartner Peer Insights for DriveNets as “attempted but inaccessible / not found via accessible pages.” (gartner.com)
• TrustRadius: no DriveNets product page or reviews were found in accessible sources during this research session (attempted via multiple TrustRadius searches). The TrustRadius homepage is accessible, but DriveNets-specific review content was not located. (trustradius.com)

Positive feedback (quotes)

• G2 (single reviewer) praised usability and modular architecture:
  • “The ease of use of the platform and its modular network architecture make it more flexible at the business level.” (G2 user, Apr 10, 2023) (g2.com)
• Vendor website includes partner/customer testimonial snippets (marketing-selected; not independent):
  • Digi Group executive quote praising responsiveness and agility vs traditional routers. (drivenets.com)

Negative feedback / criticisms (quotes)

• G2 (same single reviewer) cited:
  • “Some stability issues… network connectivity interruptions or performance issues after the implementation…”
  • “The price… can be relatively high… learning curve… can be a bit steep and some training is required…” (g2.com)
• Reddit discussion on “cloud routing” shows conceptual pushback (community opinion, not an implementation review):
  • One commenter argued what DriveNets does “has nothing to do with the ‘cloud’,” reflecting skepticism about marketing language rather than product capability. (reddit.com)

Why customers chose this product over alternatives (evidence)

• AT&T deployment is widely cited as a validation point for large-scale disaggregated core routing; DriveNets’ architecture aligns with AT&T’s OCP DDC model and Broadcom merchant silicon ecosystem. (drivenets.com)
• Independent analysis frames the main decision drivers as cost and flexibility benefits of disaggregation versus proprietary chassis (Cisco/Juniper), assuming the operator can handle operational change. (techtarget.com)

Why some customers left for competitors (churn reasons)

• Not directly evidenced in accessible public reviews for DriveNets (a major gap due to scarce public end-user reviews).

Specific outcomes customers achieved (with metrics when available)

• AT&T migration metrics are discussed in third-party reporting referencing AT&T blog claims:
  • TelecomTV reported DriveNets “currently transporting more than 52%” of AT&T’s core production traffic as of Jan 2023 (per an AT&T blog post). (telecomtv.com)
• Vendor claims on scale:
  • Company overview page claims scaling from 2.4 Tbps up to 921 Tbps router cluster. (drivenets.com)
• AI networking claims (vendor and marketing collateral) mention performance metrics like job completion time improvements in some brochures; however, these are typically vendor-published and often lack methodology in public snippets. (gsma.my.site.com)

AT&T next-gen core deployment (case reference)

• DriveNets press release (Sep 28, 2020) states AT&T deployed an open disaggregated core routing platform into production with DriveNets NOS (DNOS) and merchant silicon/white-box ecosystem partners. (drivenets.com)
• Critical evaluation:
  • This is primarily vendor-issued PR with partner quotes; it indicates validation but may omit operational issues encountered.
  • Independent follow-on reporting (TelecomTV referencing AT&T blog) suggests meaningful traffic migration by Jan 2023, which supports “real deployment” credibility. (telecomtv.com)

AI fabric scaling and multi-site GPU clusters

• Network World (May 21, 2025) reports DriveNets enhancements for multi-tenancy and multi-site GPU clusters up to ~80km apart, framed as addressing power constraints and distributed cluster design. (networkworld.com)
• Critical evaluation:
  • Feature announcements are credible but performance claims should be validated with third-party benchmarks and customer references.
  • Multi-site AI fabrics add complexity (latency, failure domains, operations). Buyers should ask for reference architectures and SLOs.

Economic/TCO studies

• ACG Research published an “Economic Comparison” whitepaper (2019) arguing cost effectiveness at scale.
• Critical evaluation:
  • This is an analyst/consultancy report; methodology and assumptions should be reviewed carefully because TCO results can vary dramatically by network design, traffic growth, vendor discounts, and staffing model. (acgcc.com)

Vendor-published “Company Overview” claims

• DriveNets collateral claims engagement “with over 100 tier-1 service providers and hyperscalers” and scaling up to 921 Tbps. (drivenets.com)
• Critical evaluation:
  • “Engagement” is not the same as production deployments; buyers should request clarity (paid pilots vs POCs vs contracted deployments).

• Best fit tends to be very large organizations where network scale and capex/opex leverage justify architectural transformation:
  • Communications service providers: typically 1,000+ employees; multi-billion revenue.
  • Hyperscalers/NeoClouds: 500+ employees; significant capex budgets.
  • Large enterprises running AI clusters: often 1,000+ employees with centralized infrastructure teams.

• Telecom and cable operators (AT&T, Comcast cited in vendor/press). (drivenets.com)
• AI infrastructure builders: hyperscalers, foundation labs, NeoClouds (vendor positioning; some independent coverage). (drivenets.com)

• Decision makers:
  • Chief Network Officer / CTO Network Services (telco)
  • VP Infrastructure / Head of Backbone Engineering
  • Head of AI Infrastructure / Data Center Networking
  • Procurement/vendor management for large hardware/software contracts
• End users:
  • Network architects, routing engineers
  • SRE/NetOps teams operating orchestration and telemetry
  • Data center network engineers (AI fabric)

• Scaling core routing beyond chassis constraints
• Reducing cost per bit and avoiding proprietary router lock-in via disaggregation
• Faster innovation cadence via software model
• For AI: alleviating network bottlenecks and improving GPU utilization (as claimed by vendor; buyer should validate with benchmarks). (drivenets.com)

• You can operationalize cloud-native disaggregation (tooling, automation mindset, and willingness to change runbooks).
• You have sufficient scale that savings and agility gains offset migration and training costs.
• You want a vendor willing to take “end-to-end responsibility” across software + certified white-box designs. (drivenets.com)

• Small/medium networks without dedicated NetOps engineering depth.
• Organizations requiring fully transparent public pricing and self-serve trials.
• Teams that cannot tolerate multi-year parallel migrations or major operational change.

• DriveNets sells a disaggregated, cloud-native networking platform that runs network functions (routing and AI back-end fabrics) as software on merchant-silicon “white-box” hardware, aiming to deliver carrier-grade scale with more flexible operations than integrated router chassis.
• The portfolio spans:
  • DriveNets Network Cloud (service-provider/core routing + broader IP/MPLS and related use cases).
  • DriveNets AI Fabric / Network Cloud-AI (Ethernet-based AI networking fabric positioned as a high-performance alternative to InfiniBand, plus orchestration and deployment services).

• Primary: Tier-1/Tier-2 communications service providers and large-scale network operators pursuing router disaggregation (core/aggregation/peering/DCI/edge). (drivenets.com)
• Secondary/expanding: Hyperscalers, “NeoCloud” GPU-as-a-Service providers, foundation model labs, and enterprises building large GPU clusters needing high-performance Ethernet fabrics. (drivenets.com)

• HQ: Ra’anana, Israel (also has a US office in Red Bank, NJ per the company website footer/contact info). (drivenets.com)
• CEO/co-founder: Ido Susan (publicly cited by DriveNets). (drivenets.com)
• Co-founder: Hillel Kobrinsky is widely cited in press coverage as co-founder. (techcrunch.com)
• Founding year is inconsistently reported across sources:
  • TechCrunch describes the company emerging from stealth in February 2019 and implies it was built prior, but does not fix a single “founded” year in the excerpt. (techcrunch.com)
  • Network World states DriveNets “got its start back in 2015.” (networkworld.com)
  • A GSMA document lists “Founded: 2016.” (gsma.my.site.com)
  • Takeaway: the market commonly places founding around 2015–2016; buyers should confirm with DriveNets for legal/vendor qualification.

• Funding (recent): DriveNets announced a $410M Series D on June 1, 2026, stating it has raised $1B total capital. (drivenets.com)
• The company claims:
  • “more than $1B in secured business”
  • “cash-flow positive since 2025”
  • Use of proceeds includes inventory build-out for AI Ethernet fabric demand. (drivenets.com)
• Prior funding coverage includes:
  • $208M Series B (Jan 2021), valuing it over $1B post-money (TechCrunch). (techcrunch.com)
  • $262M funding round (Aug 2022) (TechCrunch tag and article). (techcrunch.com)
• Customer footprint (public names mentioned across sources): AT&T, Comcast, Orange, Telefonica, KDDI appear in vendor or independent coverage. (drivenets.com)
• “Over 30% of total U.S. internet traffic” appears on the DriveNets G2 profile as a vendor claim (not independently validated in the G2 snippet). (g2.com)
• Valuation reporting: some third-party writeups cite an ~$8.5B valuation for the Series D, but this appears to be media interpretation/secondary reporting rather than stated in DriveNets’ press release (DriveNets did not publish a valuation figure in the release excerpt). Treat as unconfirmed unless verified in primary financial reporting. (siliconreport.com)

• DriveNets does not publish public list pricing or tiers on its website for Network Cloud or AI Fabric (sales-led/enterprise-style pricing is implied).
• G2 listing exists but does not provide public pricing in the captured excerpt; it is also based on only one review, limiting confidence. (g2.com)
• Practical implication: expect bespoke quotes tied to cluster size/throughput, hardware bundle choices (DriveNets-supplied vs customer-sourced ODM white boxes), support, and services.

• Likely cost drivers for buyers:
  • Hardware procurement strategy
    • DriveNets positions the platform as running on white-box hardware from DriveNets and/or ODMs (e.g., UfiSpace/Edgecore/Accton referenced by DriveNets). This can reduce hardware unit costs vs integrated chassis, but shifts responsibility to integration/testing unless the vendor provides “end-to-end responsibility.” (drivenets.com)
  • Migration/parallel run complexity
    • Independent coverage (TechTarget) describes a “step-by-step” approach: run DriveNets in parallel to absorb growth before full replacement, implying multi-year migration and dual-run operational overhead. (techtarget.com)
  • Orchestration/automation deployment
    • DriveNets DNOR orchestration is part of the solution; operational savings depend on how deeply a provider adopts the automation model. (drivenets.com)
  • Training costs
    • DriveNets offers certification/training (public one-pager), suggesting non-trivial enablement needs for cloud-native/disaggregated operations. (get.drivenets.com)
  • AI fabric: cluster bring-up and tuning services
    • DriveNets explicitly sells deployment/optimization services for AI fabrics; for buyers, this can shorten time-to-value but increases services spend. (drivenets.com)

• Traditional integrated routing vendors (service provider core): Cisco, Juniper.
• Disaggregated routing/NOS competitors: Arrcus (ArcOS), Volta Networks (mentioned in industry comparisons).
• AI Ethernet fabrics: NVIDIA Spectrum-X (Ethernet for AI), Broadcom-based open Ethernet fabrics sold via OEM/ODM ecosystems.
• Open NOS ecosystem alternative (often for DC switching rather than SP core): SONiC (appears in competitive context in the broader “open networking” market; not a direct like-for-like for SP core routing).