IBM
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 15
- IBM to ship 0.7nm Nanostack design kits in Q1 2027
IBM said it will make process development kits (PDKs) for its 0.7-nanometer "Nanostack" chip manufacturing technology available in the first quarter of 2027, letting chip designers such as NVIDIA and AMD evaluate the process before broad adoption. According to Wccftech, IBM introduced the technology in June, and the kits will include design rule manuals, design rule checking and scripts for generating layouts via EDA tools from the likes of Cadence and Synopsys. Nanostack vertically stacks and staggers transistors so n-type and p-type devices can be optimized separately; IBM claims the 7-angstrom node also delivers 40% SRAM scaling, notable because on-die SRAM takes up a large share of AI chip area.
Wccftech · 🔥 5 - IBM releases Granite Switch 4.1, combining chat, decisions and speech in one endpoint
IBM published a Hugging Face blog post titled "Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint," introducing Granite Switch 4.1. The post presents the offering as delivering chat, decision-making and speech capabilities through a single endpoint. Only the title was available in the source material, so model size, benchmarks, licensing and availability details were not disclosed.
Hugging Face · 🔥 5 - IBM links agent safety to financial oversight in its enterprise AI operating model
IBM Automation Portfolio outbound CTO Tyler Lynch told theCUBE that enterprises run very different kinds of agents — application-like, autonomous, delegated and local co-work agents such as coding agents — and that authorization, logging and telemetry practices differ across them with no settled answer yet. IBM supports Nvidia's Open Agent Safety Platform with capabilities spanning agent identity, credentials and hybrid infrastructure, and is discussing with Apptio how to extend agent safety into financial oversight, for example letting policies evaluate token telemetry and revoke an agent's API key to pause it. HashiCorp's Infragraph is now available on a limited basis to eligible Terraform customers, and IBM wants to extend it into a knowledge graph for dependency impact analysis while tying AI spending to business outcomes rather than token consumption alone.
SiliconANGLE AI · 🔥 4 - IBM survey: 80% of EMEA firms saw AI incidents reach the board, few update governance
An IBM survey of more than 3,000 business leaders across EMEA found that 80% said an AI-related incident — such as an agent failure, data exposure, AI-driven decision error or AI security breach — had reached their board or C-suite in the past year, yet fewer than half (40%) said they would change their AI governance policy or frameworks in response. The study also found 98% of executives have formally assigned AI risk responsibility at board level, including 26% with a dedicated board AI committee, while only 37% believe their governance is designed to scale with agentic AI deployment. Respondents were split on liability: 33% cited formal legal or contractual frameworks defining AI vendor liability for agent failures, 32% defined human escalation thresholds, 30% real-time monitoring and audit trails of agent decisions, and 29% the ability to immediately halt agent activity enterprise-wide. IBM's Ana Paula Assis said the AI debate has moved from adoption to governance and accountability, which must be built into how agents operate.
TechRadar · 🔥 3 - Kyndryl to run AI agents inside Toyota Finance's live IT operations
Kyndryl and Toyota Systems said they will run AI agents inside Toyota Finance's production IT operations, applying Kyndryl's agentic AI framework to change management, infrastructure design and impact analysis, working from Toyota Finance's own rules, runbooks and operational schedules to flag inconsistencies and risks. The companies framed the key point as the agents operating in the production environment rather than a pilot, but disclosed no contract value, timeline or measured results. The report also cites AWS's agentic AI lead saying almost 90% of prototypes built by Amazon teams never reached production, and a survey of 685 CIOs in which 81% said they lacked full oversight of agents built outside formal channels.
The Next Web · 🔥 0 - NordLayer launches AI Scout to track and police workplace 'shadow AI'
NordLayer announced AI Scout, a standalone product that monitors which AI services employees actually use at work, distinguishes sanctioned from unsanctioned tools, and tracks browser and desktop interactions. It can detect, redact or block potentially sensitive prompts before submission, estimates token consumption across deployed LLMs, and lets administrators inspect MCP infrastructure linked to coding agents and the capabilities available through it. NordLayer executives describe the situation as an "AI zoo" of unsanctioned LLMs and argue governance needs unified telemetry; the company cites a Writer survey finding 59% of firms spend over $1 million a year on AI, a University of Melbourne report cited by NordLayer saying 42% of employees use approved AI systems while 70% use free models at work, and IBM data putting the added cost of shadow-AI breaches at $400,000 on top of a $4.99 million global average.
TechRadar · 🔥 0 - AI doubters grow louder as IBM bets on commodity models and inference shift
On the latest theCUBE Pod, SiliconANGLE's John Furrier and Dave Vellante discussed rising public doubt about AI against reports that OpenAI's annualized net revenue is about $50 billion — roughly $20 billion below earlier figures — with shares of Nvidia, CoreWeave, Oracle and Nebius falling. The analysts called the backlash an “AI freakout moment” and said IBM is betting on model commoditization, building on Red Hat's open-source foundation to offer a four-part architecture of agents, data, automation and hybrid infrastructure plus an orchestration layer that can integrate agents from multiple vendors. The episode also cited Marvell's target of $70 billion to $90 billion in revenue for fiscal 2031 and a workload mix that has moved from about 90/10 training-to-inference a year ago to roughly 60/40, with an expected flip to 10/90.
SiliconANGLE AI · 🔥 0 - torch-spyre moves its compiler onto PyTorch Inductor
A PyTorch blog post explains that torch-spyre, the PyTorch integration for IBM's Spyre accelerator, has moved its torch.compile backend from sendnn — a compilation stack built entirely from scratch — to backend=inductor, built by extending PyTorch's own open-source Inductor. Most of Inductor carried over untouched, with the work concentrated in three hardware differences: Spyre moves and computes on 128-byte sticks rather than individual elements; each of its 32 cores has only a 2MB local scratchpad with no hardware-managed caching, so every data movement must be an explicit compiler-emitted instruction; and the cores run as a static dataflow machine, so work division must be decided at compile time. The authors say the resulting compiler is roughly 7x smaller than what it replaced, and report that changing input and output tensor tiling cut batched matmul kernel time by 29–69% across batch sizes and shapes when tensors were already resident in device memory. The Spyre passes currently hook Inductor by monkey-patching GraphLowering.codegen(), and the team is working with upstream PyTorch to formalize those extension points.
PyTorch Blog · 🔥 0 - The real AI adoption blocker: workers who can't or won't use it
TechNews cites Forbes columnist Joe McKendrick's argument that AI adoption succeeds or fails less on technology than on employee acceptance and whether AI actually fits into everyday workflows. A Resume Now survey of 1,000 employed Americans found 42% lacked confidence about integrating AI into their workflows, 44% did not know how to build AI skills, and 60% said their jobs did not require AI or gave unclear expectations. A paper from the CIVIC-AI 2026 workshop held in Singapore in July 2026 adds that most studies measure only automation capability and adoption rates, overlooking verification, exception handling and remediation costs — and the possible deskilling that follows structural job changes; Andy Thurai, founder of The Field CTO and former IBM chief strategist, also criticized token usage as a crude value metric, saying firms should disclose verification and correction costs before publishing productivity numbers.
科技新报 · 🔥 0 - AI agents crack old software as decompilation milestones hit games and COBOL
In recent weeks, matching decompilation projects for GoldenEye 007, Super Smash Bros. Melee and Halo: Combat Evolved hit milestones, and one developer reported that a fleet of 17 Claude Code agents made the 2009 Call of Duty: Modern Warfare 2 fully playable, consuming roughly 600–700 billion tokens over three months. The article credits matching decompilation's unambiguous pass/fail check, which lets coding agents loop unattended for hours, along with maturing tools such as Ghidra, objdiff, N64Recomp and ReXGlue. The same pattern is reaching enterprise legacy code: Anthropic claimed Claude can automate COBOL analysis work, IBM countered that translating COBOL is not the same as modernizing an enterprise system, and Gartner warns over 70% of mainframe migrations started this year will fail.
TechSpot · 🔥 0 - King Slide's AI server rail share nears 80%, Q2 gross margin 87.42%
A feature republished with permission from Business Today says Taiwan-listed King Slide Works, which has made only slides and rails for 40 years, holds over 30% of the global server-rail market and close to 80% in AI servers; its second-quarter gross margin was 87.42%, above the 67.7% cited for TSMC, with first-half EPS of NT$110.96. The article describes the fully in-house production line at its Chuanyi plant in Kaohsiung, roughly 3,800 accumulated patents, and rails containing more than 100 parts that must handle 40–200 kg loads plus quick release, drop prevention, damping, locking and liquid-cooling quick connects, with new ideas proposed in 3–4 days and mass production possible the following month. It adds that cash and cash equivalents reached NT$35.123 billion as of August, over 70% of total assets, and that the company has a Houston plant.
科技新报 · 🔥 0 - IBM ties enterprise AI orchestration to production readiness
Bruno Aziza, IBM group VP of software, data, AI, automation and security, told theCUBE that enterprise AI orchestration has to connect agents to the data, applications and infrastructure companies already run, with continuous visibility and oversight — warning that employee-built agents will outpace what IT can manage. He said IBM's AI operating model links agent orchestration, data integration, hybrid operations and sovereignty, while its Sovereign Core product maps more than 200 compliance frameworks to controls. Those deployment questions will be discussed at IBM TechXchange 2026, held Oct. 26–29 in Atlanta, with customer examples from Citigroup, CVS Health and DoorDash.
SiliconANGLE AI · 🔥 0 - AI-generated bug reports overwhelm open source security, IBM says AI can help fix it
Speaking at the Linux Foundation's Open Source Summit, IBM chief client innovation officer for enterprise security Jamie Thomas warned of a "tsunami in vulnerability disclosures," with roughly 66,000 unique entries expected in 2026 — about a fourfold increase over seven years ago — while time-to-exploit has shrunk from days to as little as 29 minutes. At the same time, AI-generated reports that are duplicated, low-quality or outright fake are swamping maintainers: curl shut down its HackerOne bug bounty, Google suspended its open source vulnerability rewards program, and Linus Torvalds called the Linux security mailing list "almost entirely unmanageable." Thomas argued the community should keep AI-powered vulnerability discovery but use AI to filter duplicates and bogus reports, assess severity and assist remediation, with efforts such as OpenSSF reducing the burden on maintainers.
TechRadar · 🔥 0 - Kyndryl survey: only 10% of IT chiefs use agentic AI for legacy modernization
A Kyndryl survey of 2,000 senior IT decision-makers found that just 10% of technology chiefs apply agentic AI to modernization work, four in ten enterprises still run mainframes, and 18% say their modernization efforts delivered limited or unclear value. Andy Thurai, founder of The Field CTO and a former IBM chief strategist, said vendors pitch AI to IT leaders as a cure-all but "the economics don't hold up" — and three in ten respondents already worry AI-driven modernization will bring wildly unpredictable compute costs.
ZDNET AI · 🔥 0 - torch-spyre brings IBM Spyre in as a native PyTorch device
IBM's torch-spyre team published a PyTorch blog post describing how it made Spyre, IBM's dataflow AI accelerator for inference, a native PyTorch device. The integration uses PrivateUse1 registration to give Spyre a device identity named "spyre", then maps PyTorch's allocator, stream, event and Inductor compiler abstractions onto Spyre runtime constraints such as ordered queues of typed operations, separate compute and data-movement pipelines, and ahead-of-time compiled programs with fixed layouts. IBM says this gives one path for eager and compiled execution, keeps tensors resident on device="spyre", and cuts launch overhead; no performance or benchmark figures are given, and the source text is truncated near the end.
PyTorch Blog · 🔥 0
Experience and discussion from the community
Share my IBM experienceAsk about IBM
Nobody has shared their experience with IBM yet.