AI Enterprise

IBM's chief executive says open weight AI models have become irresistibly good and cheap to run on-premises, arguing companies no longer need to rely solely on cloud providers for capable AI. The remarks fuel ongoing debate about whether enterprises will shift AI workloads in-house as open models close the gap with proprietary ones.

Google has announced Gemini 4 Carbon, a new version of its Gemini AI model, claiming it tops industry benchmarks in cybersecurity tasks. The announcement, circulating widely on social media, is drawing attention from security researchers and AI watchers discussing what the claimed benchmark leads mean for enterprise security tooling and the competitive AI landscape.

Oracle's chief executive has acknowledged that the company's enterprise software can take years to fully deploy, a notable admission as Oracle races to position itself as a major player in cloud AI infrastructure. The remark is drawing attention because lengthy rollouts clash with the fast pace of the AI boom, where customers expect rapid access to computing power and services.

Nvidia has discussed charging $2,000 per GPU annually for software used to run AI agents on its infrastructure. The company already charges more than twice that amount for its AI Enterprise software platform, so the proposed fee would represent a lower tier for agent workloads. The discussion has drawn attention from developers and industry watchers concerned about growing licensing costs on top of expensive hardware.

A Y Combinator-backed startup has raised $108 million to tackle a major bottleneck holding back enterprise AI adoption. A behind-the-scenes look at 24 hours inside the company is drawing attention, with tens of thousands engaging with the story. The funding round highlights continued investor appetite for tools that help large companies deploy AI at scale, and viewers are discussing what the startup's approach might mean for the crowded enterprise AI market.

Cognitum One has announced MetaHarness, an open-source tool for building branded, governed AI agent harnesses, timed to coincide with GitHub Universe 2026. The company positions it as a 'factory' that lets organisations deploy AI agents under their own brand while keeping control and governance policies in place. Details beyond the announcement remain limited.

OpenAI has announced that its AI models will be offered on Google Cloud, a notable move given the two companies' rivalry in artificial intelligence. The partnership means Google Cloud customers will be able to access OpenAI models alongside Google's own, expanding choice for enterprises building on cloud AI services. The announcement drew attention given the unusual competitive dynamics.

AI agents and large language model tooling are increasingly querying internal enterprise data stores directly, moving the database from a passive repository to an active execution layer where AI logic runs. Commentators in the software community warn this shift fundamentally changes where software risks hide, with security and reliability concerns now concentrating at the database layer rather than in application code.

Cisco has released an open source tool designed to track dangerous behavior from AI agents. The software monitors autonomous AI systems to flag potentially harmful actions, addressing growing safety concerns as companies deploy AI agents in production environments.

Cisco has released an open source project called Quarry, designed to monitor the behavior of AI agents. The tool aims to give organizations visibility into what autonomous AI systems are actually doing, addressing growing security and oversight concerns as companies deploy agentic AI. The announcement is drawing attention from security and enterprise software observers tracking how the industry handles AI agent accountability.

MIT Technology Review is examining how companies can unlock real value from artificial intelligence by grounding it in their own enterprise context. The argument is that generic AI models deliver limited returns, and businesses need to connect them to internal data, workflows and domain knowledge to see measurable results. The piece feeds a wider debate about moving from AI experimentation to practical enterprise deployment.

The Big Technology newsletter has released a guide to enterprise AI, laying out how companies are adopting artificial intelligence in their operations. The piece aims to explain the key tools, players and trends shaping business AI use as interest in the technology keeps climbing across industries.

A new analysis argues that artificial intelligence is changing procurement inside large organisations, altering how companies evaluate, select and purchase enterprise software. AI-driven tools can compare vendors, automate contract review and recommend products, shifting decisions away from traditional manual buying processes. The piece suggests businesses that adapt procurement to AI stand to cut costs and speed up purchasing.

Cisco has open-sourced Project Quarry, a tool designed to detect risky or unsafe behavior from AI agents. The move gives developers and security teams free access to technology for monitoring autonomous AI systems as their use spreads across enterprises. The announcement is drawing attention in software and security circles, where concerns about AI agents acting unpredictably or beyond their intended limits are growing.

Security leaders are ranking AI security as the defining priority for 2027, with every leading concern on their agenda tied to artificial intelligence in some form. The discussion covers securing AI systems themselves, defending against AI-powered attacks, and governing employee use of AI tools. Commentators say the shift signals AI has moved from an emerging risk to a core boardroom issue for security teams.

IBM has announced plans to acquire DataStax, a vendor known for its open-source data platform built around Apache Cassandra and AI capabilities for enterprises. The deal strengthens IBM's portfolio in data infrastructure and generative AI services, as the company continues a string of acquisitions targeting hybrid cloud and AI workloads. Financial terms were not disclosed in the initial report.

A reseller has listed a 300-rack AI supercomputer for sale on eBay at a price of $59.7 million, an unusual marketplace listing for hardware of this scale. The offering has drawn attention because such supercomputers are typically sold through specialised enterprise channels rather than online auction sites, and it highlights growing demand for high-end AI computing infrastructure.

Oracle announced it is integrating OpenAI's ChatGPT and Codex into its business applications while expanding its cloud chip lineup. The move ties Oracle's enterprise software more closely to generative AI tools and broadens its custom hardware offerings for cloud customers, underlining the company's push to compete in AI infrastructure and services.

Oracle is promoting AI-assisted intelligence capabilities designed to help companies get more value from the data stored in their Oracle E-Business Suite environments. The announcement highlights using artificial intelligence to analyse enterprise data and support decision-making for organisations still running the long-standing business applications suite.

Google Cloud has announced the general availability of Gemini Robotics Embodied Reasoning 2, a model aimed at reasoning for physical, embodied systems such as robots. The release makes the technology accessible to developers and enterprises through Google's cloud platform, extending Gemini's AI capabilities into robotics applications.

The University of Texas at Austin has announced UT Dell Medicine, pairing the rebranded medical enterprise with plans for a new AI-native hospital. The initiative signals a major push to embed artificial intelligence into clinical care, research and hospital operations from the ground up. Coverage in healthcare trade press frames it as a significant bet that AI-first design will shape the next generation of academic medicine.

Anthropic has released new tools that give business customers visibility into how their spending on the company's AI models is broken down and used. The move, reported by TechCrunch, comes as the frontier lab opens up more detail about usage ahead of a planned initial public offering. Observers see it as a bid for transparency with enterprise clients as scrutiny of AI costs grows.

Anthropic has launched new tools designed to show companies what their AI spending is actually delivering. The aim is to give business customers visibility into usage, costs and outcomes from their Claude deployments. The move comes as enterprises increasingly demand measurable returns on their AI investments before expanding budgets.

A startup founded by former DOGE staff says its AI agents gained access to Anthropic's business systems in less than 12 hours, according to a report by Quartz. The claim has drawn attention to how quickly autonomous AI tools can exploit security weaknesses at a major AI company, raising questions about enterprise security practices and the growing capabilities of agentic AI software.

ServiceNow has announced a rollout of secure agentic AI capabilities spanning retail, telecom, sales and IT operations. The offering is aimed at letting enterprise customers deploy autonomous AI agents with built-in security controls across core business functions. The announcement positions the enterprise software firm within the fast-moving race to commercialize agentic AI for corporate workflows, as companies weigh productivity gains against security and governance risks.

Oracle announced plans to integrate OpenAI's ChatGPT and Codex directly into its enterprise applications, while also adding Google's custom AI chips to its data center infrastructure. The moves deepen Oracle's ties to rival AI ecosystems and signal its push to compete in cloud AI services. The news was reported by Quartz and is drawing attention from tech and semiconductor watchers.

A new guide highlights the top five AI gateways for enterprise security, describing the technology as a layer that sits between AI systems and company services, functioning much like a firewall. The piece positions AI gateways as increasingly essential infrastructure for businesses deploying artificial intelligence, offering traffic control and protection as AI adoption grows across industries.

German software company SAP says it is replacing costly AI models with cheaper alternatives as part of a push to generate more than €2 billion in gains. The move signals a broader cost-discipline shift among enterprise software makers as they weigh the expense of running large AI systems against returns, and it is drawing attention from investors and tech observers watching how AI spending pays off.

Oracle has launched HeatWave ML, a capability that integrates machine learning directly into its HeatWave database service, allowing users to run ML workloads alongside analytics without moving data to separate systems. The announcement points to growing demand for built-in AI features in cloud data platforms, as vendors race to simplify how enterprises train and apply models on their stored data.

Cisco has announced that its AI Defense product is joining ServiceNow's Control Tower Alliance, a partner program built around ServiceNow's platform for managing and governing enterprise technology. The move ties Cisco's AI security tooling into ServiceNow's control framework, letting joint customers monitor and secure AI usage alongside their broader IT operations.

Microsoft used its PPCC 2026 keynote to outline its vision for the agent-native business, detailing how autonomous AI agents will be embedded across its business applications and technical platform. The talk covered both strategic direction for enterprises adopting agentic AI and the technical groundwork partners and customers can expect, drawing attention from the Dynamics 365 and Power Platform community.