Digital Plumber

Archives
Log in
Subscribe
October 20, 2026

Oct 20: NetBox Labs adds agents and ties in NVIDIA, Grafana

Digital Plumber

Plumbing the information age

Tuesday, October 20, 2026  ·  No. 128  ·  17 stories

Today's 3 things that matter

  1. NetBox Labs adds agents and ties in NVIDIA, Grafana
    NetBox becomes native to NVIDIA DSX OS and expands its source-of-truth role for AI infrastructure automation; critical for network teams scaling AI workloads that depend on accurate infrastructure data and integrated observability.
  2. Attackers exploit PAN-OS flaws to reach vCenter management
    Firewall management interfaces and VPN gateways are active attack vectors in production; both flaws demand immediate patching and restrict management plane access to trusted IPs only.
  3. OpenTelemetry graduates and adds continuous profiling as fourth signal
    OTLP is now cemented as industry standard post-graduation, eliminating vendor lock-in and enabling cross-signal correlation essential for meaningful AIOps without noise. Unified telemetry foundation enables NOC automation to correlate network, application, and infrastructure signals from single instrumentation.

Full stories below, grouped by topic.


Network Automation

NetBox Labs adds agents and ties in NVIDIA, Grafana

NetBox Labs · Oct 13, 2026 · Primary source

What happened: NetBox Labs announced partnerships with NVIDIA and Grafana Labs at NetBox Evolve 2026, introducing NetBox Designs, NetBox Analytics, NetBox Agents, and integrations with NVIDIA Configuration Manager (NVCM), Arista CloudVision, AWS VPC IPAM, and Cisco Nexus Dashboard. The platform now serves as the central intelligence layer for AI datacenter and enterprise infrastructure operations.

Why it matters: NetBox becomes native to NVIDIA DSX OS and expands its source-of-truth role for AI infrastructure automation; critical for network teams scaling AI workloads that depend on accurate infrastructure data and integrated observability.

NetBox Labs announced major platform expansions at its inaugural NetBox Evolve conference on October 13, 2026. NVIDIA integration makes NetBox the primary system of record for NVIDIA Configuration Manager within NVIDIA DSX OS, embedding NetBox at the center of GPU cluster fabric automation. New capabilities include NetBox Agents (pre-built agents for change review, compliance checking, data quality, and imports), NetBox Analytics (launching in November) with NetBox Visual Explorer for visualizing large-scale datacenter and network infrastructure, and new integrations with Arista CloudVision, AWS VPC IPAM, and Cisco Nexus Dashboard (available in October), plus preview integrations with Palo Alto Panorama, Tenable, and Kubernetes launching end of October. Grafana Labs partnership integrates NetBox with Grafana's observability stack. The announcement positions NetBox as the data foundation for agentic infrastructure operations—directly addressing the 2026 reality that 0% of enterprises currently operate with agentic NetOps, though Gartner forecasts 50% will by 2030. The platform now supports both human operators and AI agents designing and running complex infrastructure.

Read the original at netboxlabs.com


AIOps & Network Observability

LogicMonitor folds Catchpoint data into governed AI operations model

BigDATAwire · Oct 15, 2026 · Industry news

What happened: LogicMonitor announced a new operating model combining Envision platform, Edwin AI, and Catchpoint digital experience data to move IT operations from reactive alerts to governed AI-driven investigation and remediation across infrastructure, incident investigation, and automated action.

Why it matters: Organizations using Edwin AI achieved 313% ROI; the platform enables NOC and SRE teams to reduce alert noise by up to 97% through ML-driven correlation and automate triage across fragmented monitoring stacks.

LogicMonitor's Autonomous IT architecture consolidates five areas: enterprise data foundation, unified context, AI-led governed action, embedded experience, and trust/governance controls. Edwin AI applies unsupervised ML to correlate alerts, detect anomalies, identify root causes, and suppress redundant alert floods. The platform integrates with MCP for compatible AI clients, enabling agentic operations while maintaining human oversight. This represents a fundamental shift in NOC practice from reactive alert response to proactive incident investigation with bounded autonomous action—critical for teams managing complex hybrid and multi-cloud estates at scale.

Read the original at hpcwire.com

OpenTelemetry graduates and adds continuous profiling as fourth signal

DEV Community · Oct 17, 2026 · Analysis

What happened: OpenTelemetry achieved CNCF graduation in May 2026; Kubernetes Attributes Processor reached v1.0.0 in September; continuous profiling (eBPF-based) entered alpha with GA target Q3/Q4 2026. GenAI Semantic Conventions became de facto standard for AI observability, enabling hallucination rates and model quality metrics alongside traditional SRE golden signals.

Why it matters: OTLP is now cemented as industry standard post-graduation, eliminating vendor lock-in and enabling cross-signal correlation essential for meaningful AIOps without noise. Unified telemetry foundation enables NOC automation to correlate network, application, and infrastructure signals from single instrumentation.

OpenTelemetry graduated CNCF in May 2026, ranking second only to Kubernetes in project velocity with 12,000+ contributors from 2,800+ companies. Traces, metrics, and logs are now stable across every major language SDK; continuous profiling (via Parca, Pyroscope, Grafana Beyla) is increasingly integrated into OTel pipeline, providing code-level visibility alongside business metrics. GenAI semantic conventions moved to dedicated specification mid-2026 and are now universally supported. For network and infrastructure observability, OTel graduation removes vendor lock-in and enables consistent instrumentation across network devices, applications, and infrastructure—allowing NOC teams to correlate events from multiple sources (NetFlow, syslog, traces, metrics) under single trace context for unified root cause analysis.

Read the original at dev.to


Data Center & AI Networking

Micas brings co-packaged optics switches beyond the hyperscalers

Yahoo Finance · Oct 13, 2026 · Industry news

What happened: On October 13, Micas Networks announced three new switches built on Broadcom's Tomahawk 6 chip, including a compact co-packaged optics switch using Broadcom Tomahawk 6-Davisson, with production expected in early 2027. This brings CPO technology to independent vendors—previously limited to hyperscalers.

Why it matters: Co-packaged optics technology (70% power reduction versus pluggables) moves from hyperscaler-exclusive to independent vendor availability; enables mid-market and neocloud operators to build 102.4T fabrics without custom ASICs, reducing capex and time-to-market for AI clusters.

Davisson can move 102.4 terabits of data per second, which Broadcom says was double any other CPO switch available when it launched in October 2025. CPO integration reduces optical interconnect power consumption approximately 70% relative to traditional pluggables, resulting in significant TCO and sustainability wins for hyperscale and AI data centers. Micas' announcement is operationally significant because it breaks the pattern where only Meta, NVIDIA, and other hyperscalers could deploy CPO: now mid-sized data center operators and AI neoclouds can order Tomahawk 6-based systems from an independent ODM and get co-packaged optics without engineering their own switch silicon. Production timing (early 2027) aligns with the next wave of neocloud expansion and supports operators who want to reduce per-port optical power without building custom hardware.

Read the original at finance.yahoo.com


Routing & Internet

Malicious bots moved to other hosts during Yandex outage

Cloudflare Blog · Oct 14, 2026 · Primary source

What happened: Following Ukrainian drone strikes on three Yandex data centers between October 7-11, Cloudflare analyzed traffic telemetry and found that while search crawler traffic collapsed 99% and cloud services 90%, malicious bot automation did not disappear—it migrated to alternative Russian hosting providers with identical bot fingerprints and continued operating.

Why it matters: Demonstrates how BGP-observable infrastructure outages route malicious automation elsewhere rather than stopping it; shows the limits of traffic-based inference for operational security posture.

Cloudflare separated traffic from Yandex's two ASNs to clarify what actually happened. AS13238 (Yandex Search) lost 99% of traffic—almost entirely verified search crawlers indexing content. AS200350 (Yandex.Cloud) dropped 90%, but the key finding was that bot requests to gambling customer accounts fell to zero from Yandex infrastructure immediately after October 8, then reappeared with identical request signatures from Selectel and Radist ASNs. This showed the automation moved, not ceased. Using Cloudflare Radar and Bot Management data, the analysis separated legitimate indexing (93.6% of Yandex Search bot traffic came from verified crawlers) from malicious automation that adapted to infrastructure disruption. The incident highlights how AS-level visibility reveals traffic patterns but can mislead when analyzed in aggregate—a critical distinction for network operations teams tracking bot infrastructure and understanding how routing changes propagate across autonomous systems.

Read the original at blog.cloudflare.com

Fastly launches layer 3 and 4 DDoS filtering service

Fastly Blog · Oct 14, 2026 · Primary source

What happened: Fastly launched Network Protect, a Layer 3/4 DDoS defense service that filters unwanted traffic at the edge using the company's 622 Tbps global network capacity—nearly 20 times the largest recorded DDoS attack—with rule-based filtering and planned integration of an Adaptive Threat Engine for automated volumetric attack mitigation.

Why it matters: Adds edge-based L3/4 DDoS filtering to Fastly's platform with global scale and zero traffic detours; operationally relevant for teams managing multi-CDN strategies and volumetric attack resilience.

Network Protect uses Fastly's full 622 Tbps network capacity to absorb and filter layer 3 and 4 attacks inline, before traffic reaches origin infrastructure. The service launches with rule-based network traffic filtering and is designed to evolve with network-layer variants of the Adaptive Threat Engine, Fastly's automated mitigation system. The scale comparison is concrete: Fastly's capacity (622 Tbps) dwarfs Cloudflare (534 Tbps), AWS (100+ Tbps), and the largest recorded DDoS attack (roughly 30 Tbps). Fastly noted the product is the beginning of a broader vision including enhanced telemetry-driven automated mitigation. For SRE and NetOps teams, this represents an infrastructure choice point: filtering at the CDN edge versus origin-based mitigation, with clear throughput implications for attack scenarios and normal traffic patterns. The product roadmap includes rule-based filtering today with AI-driven decisions coming later.

Read the original at fastly.com


Network Security

Attackers exploit PAN-OS flaws to reach vCenter management

Palo Alto Unit 42 · Oct 19, 2026 · Primary source

What happened: Palo Alto disclosed two critical PAN-OS flaws: CVE-2026-0322 (CVSS 8.6 RCE requiring credentials), actively exploited targeting PA/VM-Series and Panorama appliances; CVE-2026-0257 (CVSS 7.8 auth bypass in GlobalProtect) where ransomware groups are bypassing MFA and pivoting directly to vCenter management. Updated threat intelligence shows attackers moving from gateway compromise to virtualization infrastructure.

Why it matters: Firewall management interfaces and VPN gateways are active attack vectors in production; both flaws demand immediate patching and restrict management plane access to trusted IPs only.

Unit 42's October 19 update to the CVE-2026-0257 threat brief reveals new attack tactics: Qilin and Settra ransomware groups are successfully bypassing MFA on GlobalProtect VPN endpoints, establishing initial foothold, then directly targeting vCenter. The flaw affects PAN-OS 10.2.0, 11.1.0, 11.2.0, and 12.1.0, plus Prisma Access versions 10.2.0 and 11.2.0. For CVE-2026-0322, attackers with valid credentials and network access to the management interface achieve root code execution on PA-Series, VM-Series, and Panorama appliances. CISA added both to the Known Exploited Vulnerabilities catalog; federal agencies must remediate CVE-2026-0322 by October 17. Organizations should apply vendor patches immediately, restrict management interface access to internal IP ranges only, and implement network segmentation to isolate firewall management traffic. The escalation from initial VPN access to hypervisor compromise reflects a material change in post-breach adversary behavior beyond simple lateral movement.

Read the original at unit42.paloaltonetworks.com

Cisco email gateway flaw exploited through crafted attachments

Cisco Security Advisory + SecurityWeek · Oct 14, 2026 · Primary source

What happened: Cisco confirmed active exploitation of CVE-2026-76519, a critical arbitrary file write vulnerability (CVSS 9.8) in Secure Email Gateway's Advanced Malware Protection (AMP) when File Reputation service is enabled and Content Scanner Tools is older than 25.3.0.5525. Unauthenticated attackers send crafted email attachments, write arbitrary files, and execute commands as root on physical and virtual appliances.

Why it matters: Email gateways are security infrastructure perimeter devices; root compromise exposes credentials, cryptographic material, and provides pivot path to internal networks. Requires immediate Content Scanner Tools update or AsyncOS version upgrade.

Cisco discovered the flaw internally during security scanning; exploitation became active in October 2026. The vulnerability stems from improper handling of email attachments when AMP's File Reputation service processes them. Attack path is simple: malicious attachment bypasses file analysis when scanner tools are outdated, leading to file write and root execution. CISA added CVE-2026-76519 to Known Exploited Vulnerabilities on October 14, requiring federal agencies to patch by October 17. Cisco shipped fixes on September 14 in AsyncOS 15.5.5-014 and 16.5.0-780; organizations must upgrade all affected gateways immediately. Attackers with root access can remove local logs and evade detection; network-level logging outside the appliance is essential for forensics. For orgs on AsyncOS 16.0, the fixer for this CVE differs from the September SQL injection patch, requiring careful version verification.

Read the original at securityweek.com

Juniper patches five critical remote code execution flaws in Junos

SecurityOnline · Oct 14, 2026 · Primary source

What happened: Juniper released October 2026 security bulletins covering dozens of critical Junos OS vulnerabilities including five CVSS 9.8 flaws affecting SRX Series and Junos OS Evolved platforms: CVE-2026-102182 (MX Series certificate-based IKE RCE), CVE-2026-101954 (full device takeover via J-Web), CVE-2026-101986 (pre-authentication CGI execution), CVE-2026-102175 (service-deployment RCE/DoS), CVE-2026-33792 (GRE tunnel stack buffer overflow). CVE-2026-102084 (CVSS 8.1) allows authenticated remote code execution in ADVPN daemon via malformed attributes.

Why it matters: Firewall and routing infrastructure must be patched immediately; J-Web management interface and IKE/ADVPN services are network-facing and frequently exposed; unauthenticated remote RCE on routing platforms is critical for BGP integrity.

The five CVSS 9.8 vulnerabilities span different attack vectors: Certificate-based IKE packets trigger root-level code execution on MX Series with SPC3; unauthenticated network attackers achieve full device takeover when J-Web is enabled; pre-authentication CGI execution via J-Web; service-deployment processing flaws enable RCE or DoS; and GRE tunnel processing allows stack buffer overflow for root execution. CVE-2026-102084 (CVSS 8.1) is particularly notable for ADVPN deployments—requires authenticated access but allows complete device control if an attacker has compromised a peer or acquired the shared secret. None are currently exploited in the wild. Fixed versions vary by CVE and Junos branch (23.2R2-S7.7, 23.4R2-S8.13, 24.2R2-S5.5 are common fix points). Organizations should upgrade SRX Series and MX Series platforms immediately and validate J-Web is disabled unless explicitly required for management. ADVPN users should audit peer authentication and shared secret management.

Read the original at securityonline.info


Startups & Funding

Boomi buys Elementary Data for data pipeline observability

Boomi · Oct 19, 2026 · Vendor release

What happened: Boomi announced acquisition of Elementary Data for approximately $40-45 million in cash and stock, adding data observability, quality, and governance capabilities to the Boomi Enterprise Platform. Elementary, founded in 2022 as an open-source project, is now adopted by 1,500+ data teams with 1.1M+ monthly downloads.

Why it matters: As enterprises operationalize AI agents and workflows, data quality and governance become critical bottlenecks; Elementary's observability tools flag data anomalies before they reach models or automated systems, directly improving AI reliability.

Elementary provides unified observability, quality, governance, and discovery tools for data pipelines. Founded in 2022 as an open-source project, it achieved adoption by 1,500+ data teams with 1.1M+ monthly downloads and a ~4,000-member community before launching Elementary Cloud in 2024 with enterprise-grade capabilities. Elementary builds software that monitors data quality and reliability moving through enterprise pipelines, flagging anomalies before they reach dashboards, models or automated systems. The 17-person team, all based in Israel, joins Boomi; this is Boomi's third Israeli acquisition. Following the Lunar.dev acquisition, Elementary strengthens the Boomi Enterprise Platform with observability capabilities enterprises need to deliver trusted data for AI agents. Boomi confirmed no immediate changes to the open-source repository, release process or community use following the acquisition.

Read the original at boomi.com


People Moves

Prevalent AI names JD Sherry chief revenue officer

Prevalent AI · Oct 15, 2026 · Primary source

What happened: Prevalent AI appointed JD Sherry as Chief Revenue Officer, leading global go-to-market organization across sales, marketing, and partnerships. Sherry joins from ISTARI, where he was Managing Partner and Global Head of Sales advising boards on cyber resilience.

Why it matters: CRO appointment supports Prevalent AI's expansion beyond cybersecurity into wider AI context platform market following $22 million growth investment.

Prevalent AI appointed JD Sherry as Chief Revenue Officer, leading the company's global go-to-market organization across sales, marketing, and partnerships. Sherry joins from ISTARI, where he was Managing Partner and Global Head of Sales advising executive leadership teams and boards on cyber resilience. He brings nearly three decades of cybersecurity experience and has served on boards of several ISTARI portfolio companies, including Prevalent AI since 2021. Prior to ISTARI, Sherry was Chief Strategy Officer and CRO at Remediant, Inc., and held leadership roles at Optiv, Cavirin and Trend Micro. The appointment follows Prevalent AI's recent $22 million growth investment from Integrity Growth Partners, which is funding global expansion, deeper leadership team, and extension of its Data Fabric and Knowledge Graph beyond cybersecurity into wider use cases requiring trusted enterprise context.

Read the original at globenewswire.com


Telco & Cable AI

AT&T and Verizon run AI inside radio network optimisation

VoIP.review · Oct 20, 2026 · Industry news

What happened: AT&T uses AI across planning, forecasting, deployment, and network optimization, with trials with Ericsson delivering significant improvements in throughput and spectral efficiency. Verizon combines centralized rApps for slower network intelligence tasks with local AI inside the gNodeB for faster decisions, achieving better downlink spectral efficiency through AI-based link adaptation.

Why it matters: Both carriers report field-validated performance gains from AI-RAN tools, signaling move from pilots to production deployment that will shape network automation strategies for operators.

Operators are moving artificial intelligence deeper into the radio access network, with AT&T and Verizon reporting real performance gains from AI-based RAN tools shared at RCR Wireless News's Telco AI Forum 2026. AT&T's RAN technology team outlines current AI use cases across planning, forecasting, deployment, and network optimization, applying rApps across parts of the network lifecycle to support automation and help engineers manage growing network complexity. AT&T trials with Ericsson delivered significant improvements in throughput and spectral efficiency—gains that matter because spectrum remains costly and limited. Verizon is combining two approaches: centralized rApps handle slower network intelligence tasks while local AI inside the gNodeB supports faster decisions near the radio site. Verizon saw better downlink spectral efficiency through AI-based link adaptation, and another field trial also improved coverage, pointing to stronger user experiences without always adding new sites. Verizon wants AI features to beat current rule-based methods—rule-based systems follow fixed instructions, while AI can adjust from patterns. Both operators emphasize the transition from experimental deployments to production use, setting expectations that field-validated AI-RAN is now the reference architecture rather than emerging research.

Read the original at voip.review

Brownfield operators accelerate non-real-time RIC rollouts across cell sites

IEEE ComSoc Blog · Oct 19, 2026 · Research

What happened: SMO/Non-RT RIC adoption is accelerating among brownfield operators including AT&T, Verizon, Telus, Vodafone, DT, Swisscom, STC, NTT DoCoMo, Chunghwa Telecom and Telstra to enable network-wide automation across thousands of cell sites. Annual spending on AI-RAN automation and optimization software and services is expected to grow at CAGR of 9% over the next four years, reaching $1.1 billion by end of 2030.

Why it matters: Documents concrete RAN automation adoption across major carriers globally, confirming SMO/RIC as enabling architecture and providing market sizing for network ops teams evaluating AI-RAN automation roadmaps.

SMO/Non-RT RIC adoption is accelerating among brownfield operators to enable network-wide automation across thousands of cell sites through vendor-supplied, third-party and internally developed rApps, with RAN vendors and application developers including Nokia, NVIDIA, Radisys, IS-Wireless, Cohere Technologies, DeepSig and others already committed to support the E3 interface. An alternative path vendors are pursuing is proprietary AI-enabled, real-time optimization capabilities as software upgrades to existing RAN platforms through subscription-based licensing models. The report identifies two distinct deployment strategies: centralized orchestration via SMO/RIC platforms and distributed embedded AI. AI-RAN automation market growth is largely driven by growing investments in centrally coordinated SMO/RIC-enabled automation at the network level and AI-for-RAN software upgrades for embedded real-time intelligence replacing static rule-based algorithms. The adoption curve among named carriers demonstrates that RAN automation is transitioning from selective pilots to broad enterprise deployments.

Read the original at techblog.comsoc.org

Scaleway commits three billion euros to European AI infrastructure

TelecomTV · Oct 19, 2026 · Industry news

What happened: Telcos and technology partners pressed ahead with efforts to develop sovereign AI infrastructure and autonomous network deployments. Scaleway, the cloud and AI unit of European telco Iliad Group, is investing in new infrastructure deployments in Scandinavia with €3bn committed to AI-dedicated infrastructure and platforms via its datacentre division OpCore and Scaleway.

Why it matters: European telcos are moving beyond network automation to build sovereign AI infrastructure, with material capex commitments creating new operational and procurement implications for network ops teams.

Telcos and technology partners pressed ahead with efforts to develop sovereign AI infrastructure, identify opportunities for autonomous network deployments, and bring agentic AI to bear across networks, enterprise services and SME markets, with standards bodies moving to shape the emerging AI infrastructure landscape while cybersecurity and datacentre interconnect capacity featured prominently. Scaleway, the cloud and AI unit of European telco Iliad Group, is investing in new infrastructure deployments in Scandinavia, expanding its compute capacity and positioning itself to be a key provider of digital sovereign cloud and AI services in Europe, backed by telecom and tech entrepreneur Xavier Niel and Iliad's success as an alternative telco. Iliad committed €3bn into AI-dedicated infrastructure and platforms via its datacentre division OpCore, a joint venture with InfraVia, and Scaleway. This signals a shift in European telco strategy: rather than merely automating existing networks, carriers are building AI-native infrastructure as a platform play to serve enterprise customers seeking alternatives to hyperscaler-dependent architectures.

Read the original at telecomtv.com


AI Model Providers

DeepSeek updates Flash model and opens search API

TechNode · Oct 19, 2026 · Industry news

What happened: DeepSeek released V4.1 Flash-1017 on October 17, updating its Flash model with improvements in reasoning, coding, agent tasks and visual understanding. DeepSeek's proprietary search API is now formally available to developers via API starting October 17.

Why it matters: The model offers stronger benchmark performance, faster task completion, and shorter outputs, with open MIT-licensed weights available for self-hosted deployment.

DeepSeek-V4.1-Flash-1017 is an updated version of the existing V4.1-Flash model released October 17, 2026, not a separate model family. The October 17 release improved performance across multiple benchmarks (HLE 42.6, DeepSWE 77.0, ProgramBench 29.1, Automation-Bench 60.5) while reducing task runtime and output length. The experiential improvement in agentic coding scenarios is striking, with the model trained using self-feedback reinforcement learning to enhance its ability to build complete projects from scratch, supporting "wish-based" programming tasks. The update is live across DeepSeek's web, app, and API endpoints simultaneously. Model weights are available on Hugging Face under the MIT license, with inference code and deployment instructions enabling developers to run the model on their own infrastructure.

Read the original at technode.com


AI Industry & Policy

US and EU AI rules diverge for multinational developers

Origin Brief · Oct 19, 2026 · Analysis

What happened: The US rebranded its AI safety apparatus with an explicit mandate to resist foreign regulation of American AI, while the EU activated enforcement powers. This divergence creates a compliance asymmetry that multinational AI developers must navigate: what satisfies US policy expectations may actively conflict with EU legal obligations.

Why it matters: The convergence of accelerating AI adoption, active enforcement, and public skepticism creates a governance pressure cooker. Organizations positioning AI governance as a competitive differentiator are best positioned to navigate this environment.

The US rebranded its AI safety apparatus as 'super intelligence' promotion with an explicit mandate to resist foreign regulation of American AI, while the EU activated enforcement powers and MEPs called for binding AI liability legislation. Executive Order 14434 directed NIST to rebrand AI as 'super intelligence,' renaming CAISI to CAISSI with a mandate that includes guarding against 'burdensome and unnecessary regulation of American technologies by foreign governments.' A new EU report published October 14, 2026 found that AI use has increased to 18% in firms in EU strategic sectors, more than doubling in almost every sector. This represents a fundamental shift in how regulatory frameworks are developing: the US is explicitly prioritizing competitive velocity and resisting external constraints, while the EU enforces a comprehensive statutory regime. For multinational enterprises, this means maintaining separate governance tracks—compliance strategies that work in one jurisdiction may create legal exposure in another.

Read the original at originbrief.app

Modal Labs raises 750 million dollars at 15.75 billion valuation

AI Funding Tracker · Oct 18, 2026 · Industry news

What happened: AI startups announced about $2.42 billion in primary funding across 19 rounds of $20 million or more between October 12 and October 18, 2026. Modal Labs' $750 million Series D was the largest round, at about 31% of the total. AI infrastructure and chips raised about $1.58 billion across five rounds, or 65% of the total.

Why it matters: Capital concentration in compute and inference infrastructure reveals where enterprises expect bottlenecks and differentiation in production AI systems. Nearly two-thirds of funding went to infrastructure rather than applications.

AI startups announced about $2.42 billion in primary funding across 19 rounds of $20 million or more between October 12 and October 18, 2026. Modal Labs' $750 million Series D was the largest round, at about 31% of the total. AI infrastructure and chips raised about $1.58 billion across five rounds, or 65% of the total. Modal, Infinigence AI, BlueCore Compute, Salience Labs and Mueon made up the group. The capital allocation is telling: nearly two-thirds of a $2.42B funding week went to infrastructure—compute, serverless platforms, and semiconductors—rather than applications. This reflects mounting conviction that the real margin will accrue to providers controlling inference capacity and model routing, particularly as enterprises move from pilot to production deployment. The concentration also signals investor confidence that these infrastructure plays can sustain premium valuations as model costs compress and latency becomes the differentiator.

Read the original at aifundingtracker.com


Read this edition on the web · The week in network intelligence · Vendor Radar

Digital Plumber is AI-curated and AI-summarized, with no human review before publishing. Verify before acting on anything here. How it works.

Don't miss what's next. Subscribe to Digital Plumber:
← Newer Oct 21: RPKI now protects 70% of internet routes after China surge Older → Oct 19: RPKI now protects 70% of internet routes
Powered by Buttondown, the easiest way to start and grow your newsletter.