The easy-money phase of the artificial intelligence infrastructure boom may be approaching a more selective chapter. Elon Musk has warned that smaller cloud computing providers could face “guppy-level death” as the funding that powered rapid industry expansion becomes harder to secure. Building an AI cloud requires expensive accelerators, high-density data centers, networking equipment, cooling systems, and long-term electricity supplies. Much of that spending occurs before customer revenue arrives. If borrowing costs rise, investors become cautious, or expected demand takes longer to materialize, smaller operators can quickly find themselves squeezed. Demand itself remains substantial. IDC reported that AI infrastructure spending totaled $318 billion in 2025, more than double the $153 billion recorded in 2024. The research group forecasts that annual spending will exceed $1 trillion by 2029. In Q3 2025 alone, spending reached $86 billion, while cloud and shared deployments represented more than 86% of the market. Fast market growth does not mean every supplier benefits equally. AI infrastructure is unusually capital-intensive, and the economics depend heavily on utilization. A GPU sitting idle still carries financing, depreciation, maintenance, and power-related costs. Providers therefore need a steady stream of workloads at prices high enough to recover their investment. That dynamic favors Amazon Web Services, Microsoft Azure, and Google Cloud. Together, AWS, Microsoft, and Google Cloud accounted for 68% of enterprise cloud infrastructure spending in Q4 2025, when the overall quarterly market reached $119 billion, according to Synergy Research Group. Their advantage extends beyond market share. Each can combine financing capacity, chip procurement, global data-center footprints, software ecosystems, and established enterprise sales channels. insight.tmcnet.com/insight/e…

SoftBank is reportedly considering a $50 billion technology investment program in Malaysia, a potential commitment that would sharply raise the stakes in Southeast Asia’s race to build artificial intelligence and cloud infrastructure. The proposed spending would cover AI, data centers, and other advanced technologies, according to CNBC. Potential locations include Kuching in Sarawak as well as sites across Peninsular Malaysia. For now, however, the $50 billion figure represents a reported plan rather than a finalized or completed investment. That distinction matters. Large infrastructure programs are generally deployed in stages, with construction schedules, financing, energy procurement, and customer demand influencing how much capital is ultimately committed. SoftBank’s proposal could encompass direct investment, partnerships, and computing services rather than a single data center project. Even so, the scale is striking. Google has committed $2 billion to establish its first Malaysian data center and cloud region, while Microsoft has announced $2.2 billion for cloud and AI infrastructure and related digital initiatives. Amazon Web Services plans to invest $6.2 billion in Malaysia through 2038. SoftBank’s reported figure is far larger, although the scope and implementation period may also be broader. Malaysia is not starting from scratch. Data-center-related investments reached RM385.7 billion between 2021 and the first half of 2026, giving prospective investors an expanding base of facilities, contractors, connectivity providers, and enterprise customers. insight.tmcnet.com/insight/s…

Carrefour’s recent cyber incident has sharpened attention on a risk that extends well beyond payment terminals and customer databases. For supermarkets, a cyberattack can reach refrigeration controls, warehouse systems, temperature sensors and logistics platforms. Those systems affect whether food remains safe and whether stores can keep shelves stocked. The distinction matters. A retailer might continue accepting payments while losing confidence in temperature readings from freezers or refrigerated trucks. Manipulated telemetry could indicate that products remained within approved ranges when they did not. Ransomware could make monitoring dashboards unavailable just as employees need to determine which goods can still be sold. Digital food systems increasingly rely on cold-chain monitoring and Internet of Things devices. Research published by Wiley in 2026 identified ransomware, distributed denial-of-service attacks, insecure IoT and operational technology, and supplier compromise among the exposures created by that dependence. Refrigeration equipment is no longer just physical machinery. Sensors, gateways, cloud dashboards and maintenance connections have turned it into a cyber-physical environment. The same applies to automated warehouses and transportation fleets. A compromised component can create consequences in the physical world, even if the attacker never gains access to a cash register. The pressure is particularly visible in Asia-Pacific. Akamai reported that commerce bot activity across the region increased 63% during 2025. Retail also represented 51% of API-targeted Layer 7 DDoS activity there. Those figures reflect attacks against the digital interfaces that increasingly connect shoppers, suppliers, inventory applications and store operations. insight.tmcnet.com/insight/c…

Education providers should compare network monitoring options by how well each measures user experience, wireless capacity, multivendor infrastructure, and incident response. Network monitoring is the continuous collection and analysis of availability, performance, traffic, and path data. A school network can appear healthy in the operations center while students struggle to join a lesson, faculty members lose video quality, or an assessment platform slows to a crawl. Device uptime reveals only part of network performance. According to a 2025 BUFFALO survey reported by DX Magazine, 70.9% of education-technology personnel wanted to improve school Wi-Fi, 55.7% reported instability when many devices connected simultaneously, and 44.3% experienced variable video streaming. Centralized management was already in use among 69.8% of respondents, yet license renewals and multivendor management remained major concerns for 46.7% and 38.3% of respondents, respectively. Demand is expanding beyond traditional monitoring as a result. SNS Insider valued the global network-observability market at $3.30 billion in 2025 and projects it to reach $10.65 billion by 2035. Cloud deployments represented 69.2% of the market in 2025. Education providers are no longer monitoring only switches. They support cloud applications, managed devices, guest access, streaming media, remote users, security controls, and aging infrastructure, often with a small team. insight.tmcnet.com/insight/e…

Auto dealerships should evaluate Voice over Internet Protocol (VoIP) (voice calling delivered over IP networks) by testing call routing, customer relationship management integrations, network resilience, caller authentication, analytics, and support against actual sales and service workflows. A customer calls about an advertised vehicle, gets transferred from the business development center to sales, and then repeats the same information because the customer relationship management (CRM) record did not follow the call. Later, an automated campaign sends a conflicting message. That fragmented experience is one reason dealerships are examining VoIP as a connected workflow rather than merely a replacement for desk phones. Digital Dealer reported that, among 600 U.S. dealerships, 81% had lost customer conversations or leads because CRM, chat, and inventory systems were disconnected. The same research found that 65% of respondents said those disconnections delayed follow-up. Because the research was vendor-sponsored, buyers should treat both figures as directional findings from the surveyed dealerships rather than universal industry benchmarks. The practical problem spans sales, service, parts, finance, and the business development center. Each department may use separate direct-dial numbers, call queues, voicemail boxes, Short Message Service (SMS) tools, and CRM fields. A useful requirements document maps what happens to the caller ID, customer record, recording, transcript, and assigned employee when a Session Initiation Protocol (SIP) call (an IP-based signaling method for starting and managing calls) crosses those boundaries. Cloud telephony products from RingCentral, 8x8, and Zoom Phone commonly provide routing, conferencing, mobile applications, analytics, and integrations. Those categories look similar on a comparison sheet. The differences become clearer when a dealership tests an actual service-booking call or an after-hours sales inquiry. insight.tmcnet.com/insight/h…

A virtual receptionist is a remote human, artificial intelligence agent, or blended service that answers and routes business calls. Professional-services teams should evaluate these solutions by thoroughly testing intake, scheduling, transfers, integrations, privacy, accessibility, and measurable call outcomes. A prospective client calling a law firm rarely thinks about its phone architecture. The caller simply wants a person or system capable of identifying their issue, determining its urgency, and arranging the appropriate next step. Demand for that experience is helping expand the live virtual-receptionist market. Verified Market Research valued the category at $1.64 billion in 2025 and projects it to reach $3.42 billion by 2033, representing a 9.6% compound annual growth rate. This forecast covers the virtual-receptionist service category rather than the broader business voice-over-internet-protocol or contact-center software markets. For professional-services buyers, however, the decision involves more than choosing an answering service. It requires designing a controlled workflow across voice, video, SMS, calendars, and client-management systems. A professional-services firm should begin by documenting why people call and what each call requires. Common categories include prospective-client intake, appointment changes, billing questions, document-status requests, and time-sensitive matters. Those categories require different handling rules. A receptionist might schedule routine consultations through Microsoft 365 but route urgent legal matters to an on-call queue using SIP, the voice-over-internet-protocol session standard defined in RFC 3261. A billing question might create a CRM task rather than interrupt an accountant during a client meeting. insight.tmcnet.com/insight/h…

Megaport is weighing a major capital raising as it looks to expand from connecting cloud environments and data centres into supplying the computing capacity behind artificial intelligence workloads. The reported A$827 million fully underwritten entitlement offer would be one of Australia’s largest follow-on equity raisings of 2026. The potential transaction follows a strong run for Megaport shares, which are up roughly a third this year and give the company a $3.8 billion valuation. That market support provides a window for raising equity while raising the stakes: investors would be backing a broader infrastructure strategy rather than the established virtual networking model. Megaport operates a software-defined network linking more than 1,100 data centres. Customers use that footprint to connect cloud services, facilities, and private infrastructure without relying solely on fixed, long-duration telecommunications arrangements. Megaport is now seeking to place distributed GPU capacity closer to those connected environments. According to The Australian Financial Review, Megaport’s push includes large US contracts targeting the AI infrastructure market. The company has reportedly secured four AI-infrastructure contracts worth A$458.9 million and proposed investing about A$350 million in a distributed pool of Nvidia GPUs. GPUs represent a fundamentally different operational model compared to network connections. Nvidia accelerators are expensive, power-intensive assets that can become commercially exposed if customer demand, pricing, or hardware generations shift faster than expected. Megaport must manage procurement, deployment, and utilisation while continuing to deliver reliable connectivity across its existing footprint. insight.tmcnet.com/insight/m…

The reported revenue figures put SpaceXSI’s AI operation ahead of Starlink for the quarter, highlighting how quickly commercial demand for artificial intelligence is reshaping technology portfolios. SpaceXSI generated $7.56 billion, while Starlink recorded $5.25 billion. That comparison is striking because Starlink already represents a large-scale connectivity business with substantial physical infrastructure. SpaceXSI’s larger quarter suggests that AI services can reach enormous revenue volumes even while the underlying market remains young and intensely competitive. Cloud services and Cursor are among the offerings identified with the AI business, although separate revenue contributions were not provided. The broader economics are not as simple as cheap software producing easy margins. The cost of running an individual AI query has dropped sharply, but lower prices encourage customers and software agents to submit far more queries. This is the familiar rebound effect, applied to machine intelligence. Cheaper inference does not necessarily mean a smaller infrastructure bill. Agentic systems can call a model repeatedly while planning work, gathering information, checking results, revising outputs, and coordinating with other applications. A task that appears to be one user request may trigger dozens or hundreds of model interactions behind the scenes. Deloitte projects that average inference cost could fall from $0.04 per million tokens in 2025 to roughly $0.01 by 2030, while warning that agentic workflows may expand demand faster than unit-cost reductions contain it. insight.tmcnet.com/insight/s…

A vote by a rural Nevada county’s board of commissioners has pushed an increasingly common land-use conflict into public view: whether agricultural property should make way for data centers. The immediate proceeding is local, but the underlying issue stretches well beyond Nevada. Artificial intelligence and cloud computing are driving demand for large campuses, while suitable sites near established technology hubs are becoming harder to secure. Rural land can look attractive on a site-selection spreadsheet. Nevada offers space, access to major Western markets, and potential connectivity to regional fiber and electricity networks. Land may also be assembled in larger parcels than in urban areas. For developers facing lengthy permitting and grid interconnection queues elsewhere, those attributes carry weight. Farmland, however, is not vacant land in any practical sense. It supports local production, shapes community identity, and may depend on water and infrastructure systems that were not designed for industrial computing facilities. The scale of projected electricity demand raises the stakes. Lawrence Berkeley National Laboratory found that U.S. data-center electricity use nearly tripled from 2016 to 2023. Data centers represented 4.4% of national electricity consumption in 2023 and could account for 6.7% to 12% by 2028. Planned facilities are also becoming larger, with average new sites expected to draw well over 100 MW during the next decade. That is not a routine commercial load. A single large campus can require new substations, transmission upgrades, backup generation, and additional utility planning. Who pays for that buildout? If project agreements and utility tariffs do not assign costs carefully, existing households, farms, and small businesses could shoulder part of the expense through higher rates or delayed infrastructure investment. The international trajectory points in the same direction. The International Energy Agency reported that global data-center electricity consumption rose 17% in 2025 and projected demand to roughly double from 485 TWh in 2025 to 950 TWh by 2030. Microsoft, Google, and Meta are among the companies whose hyperscale infrastructure contributes to demand for large campuses. Increasingly, those campuses are moving outward as urban land, available power, and permitting capacity tighten. insight.tmcnet.com/insight/r…

SMBs should compare IT support by responsibility, coverage, security, escalation, provider stability, and total cost. The best model depends on the capabilities the organization intends to keep in-house. IT support formerly meant fixing laptops, resetting passwords, and keeping a server running. This definition is no longer sufficient. Even relatively small organizations rely on cloud applications, remote endpoints, identity platforms, managed networks, backups, and cybersecurity controls spread across several environments. Internal IT teams are increasingly asked to contribute to automation, business analytics, integrations, and strategic planning. Spending half a day troubleshooting routine endpoint problems is not a productive use of a senior administrator's time. Managed delivery has consequently become part of mainstream IT planning. According to an IDC forecast, 32% of IT budgets will shift to managed delivery models by 2025. While this functions as an enterprise-oriented benchmark rather than an SMB-specific statistic, it indicates a broader change in how organizations obtain technical capacity. The provider market has risks of its own. Omdia's 2025 U.S. Managed Services Industry Analysis examines managed-services spending and the mix of infrastructure, cloud, and support offerings sold by MSPs. Its research found that nearly 25% of surveyed MSPs were not regularly profitable. TMCnet's coverage of the Omdia findings highlighted the ongoing tension between provider revenue growth and operational profitability. insight.tmcnet.com/insight/h…

A payment exception reaches the accounting team, but its identifying data lives across the core banking platform, a payment hub, and a spreadsheet maintained by operations. Finance cannot reconcile the transaction until IT extracts another file. Meanwhile, payroll, procurement, and regulatory reporting follow separate approval paths. That fragmentation helps explain rising interest in financial-services ERP. Mordor Intelligence's current banking and financial-services ERP forecast values the market at $10.47 billion in 2025 and forecasts it to reach $16.24 billion by 2031. Within that market, cloud ERP is projected to grow at a 7.83% compound annual growth rate through 2031. Buyers should begin by identifying the processes creating delay or control exposure. Typical candidates include manual journal entry, duplicate vendor records, payroll adjustments, intercompany reconciliation, and payment exceptions that require several systems to investigate. The scope also needs clear boundaries. ERP generally handles finance, procurement, fixed assets, payroll, and management reporting. It does not replace a core banking engine, card processor, loan-origination platform, or real-time payment switch. Treating ERP as a universal replacement can inflate cost and complicate integration. Product demonstrations tend to emphasize dashboards. Buyers gain more from end-to-end transaction visibility: an ISO 20022 payment message—a standardized format for exchanging financial data—enters the payment hub, generates fees, posts to the general ledger, and appears in a reconciled finance report. insight.tmcnet.com/insight/h…

A practical cloud operating model defines how a professional-services firm connects applications, governs data and identity, assigns operational responsibilities, controls costs, and measures service performance across public cloud, private infrastructure, and SaaS platforms. Consider a mid-market professional-services firm whose consultants work in Microsoft 365, finance teams run SAP ECC, and project data sits in an on-premises SQL Server database. Staff may copy billing details into spreadsheets, email documents for approval, and wait for IT to reconcile project records across several applications. The core problem is how data, access controls, and workflows connect. Cloud adoption introduces new integration points. Gartner forecast worldwide public-cloud end-user spending to reach $723.4 billion in 2025, including $299.1 billion for SaaS. Furthermore, Flexera's 2025 report found that surveyed organizations used an average of 2.4 public-cloud providers, with 70% relying on a hybrid-cloud strategy. For professional-services buyers, managing these disparate environments creates a practical challenge. A firm may use Microsoft Azure for identity and collaboration, Amazon Web Services (AWS) for a client analytics workload, and a SaaS practice-management platform while retaining SAP and document archives on private infrastructure. Every boundary introduces decisions about authentication, data classification, latency, and cost allocation. The evaluation should begin with a specific workflow, such as converting approved consultant time into a client invoice. Buyers map the systems involved, record formats, manual approvals, data owners, and regulatory constraints. A useful diagram might show a time-entry SaaS application sending JavaScript Object Notation (JSON) through a representational state transfer application programming interface (REST API), middleware translating the payload into an SAP Intermediate Document (IDoc), and invoice status returning to a client portal. insight.tmcnet.com/insight/h…

Barclays is moving from limited AI experimentation toward a restructuring program with explicit staffing and financial targets. The bank plans to cut 5,000 positions during 2026 while redirecting some of the resulting savings into technology and AI investments. The scale is material. Barclays employed about 93,000 people globally at the end of 2025, meaning the planned reductions represent roughly 5% of its operating base. According to BNN Bloomberg, around 60% of the affected positions are in support functions, including developer roles. That detail complicates the conventional idea that AI primarily threatens repetitive administrative work. Developers are often expected to benefit from coding assistants, automated testing and faster documentation. Yet those same capabilities can allow smaller engineering groups to maintain applications, modernize systems and produce code more quickly. Productivity tools can support technical employees while also reducing the number of positions an employer believes it needs. About 1,000 Barclays roles were already eliminated in the first half of 2026, leaving approximately 4,000 additional reductions expected by year-end, Banking Dive reported. Barclays also spent £307 million on restructuring during the first nine months of 2026. That spending covered workforce reductions, technology investment and operational redesign, showing that near-term cost cutting itself carries a meaningful price. Barclays is not presenting AI as the sole explanation for the changes. Its chief executive described AI as the "icing on the cake," placing it alongside infrastructure upgrades, process simplification and broader technology modernization. The wording matters because large banks rarely become more efficient simply by installing an AI assistant. Legacy applications, duplicated workflows, fragmented data and extensive approval chains can limit the impact of even capable models. insight.tmcnet.com/insight/b…

Apple and Meta are converging around a consequential idea for the AI industry: more computational work can happen where users interact with AI, rather than inside a distant data centre. Apple Intelligence’s on-device capabilities and Meta’s AI glasses represent different product strategies, but both test the same underlying model. Phones, wearables and other edge devices can run selected inference tasks locally, while cloud systems handle training, complex reasoning and workloads that exceed device limits. That division of labour could reshape AI infrastructure spending. It also complicates the popular assumption that each new AI feature automatically requires another large cluster of data-centre GPUs. The energy backdrop is hard to ignore. According to the International Energy Agency, global data-centre electricity demand grew 17% in 2025, while demand from AI-focused facilities rose 50%. Overall data-centre electricity consumption is projected to increase from 485 TWh in 2025 to 950 TWh in 2030. AI-focused demand is expected to triple over that period. Capital spending is rising just as quickly. Five major technology companies spent more than $400 billion in 2025, driven largely by data-centre investment, and that total is forecast to rise another 75% in 2026. insight.tmcnet.com/insight/a…

Google is backing a virtual power plant designed to turn electricity conservation by homes and businesses into a resource for an increasingly strained grid. Participants could receive payments for temporarily reducing or shifting consumption, helping utilities manage peaks as AI data centers add large, concentrated loads. The approach does not involve building a conventional generating station. A virtual power plant, or VPP, coordinates distributed assets and flexible loads through software. Those resources can include home batteries, rooftop solar systems, electric vehicles, smart thermostats, commercial building controls and industrial equipment. When the grid tightens, the operator can request or automate small adjustments across many participating sites. While one household changing its air-conditioning schedule will not transform a regional power market, thousands of coordinated devices might. Google’s interest reflects the speed at which its own energy requirements are changing. The company reported that electricity consumption increased 37% year over year in 2025, its largest annual rise, while demand climbed more than 250% from 2019. Data Center Knowledge has reported that AI growth is outpacing the decarbonization of grids serving expanding digital infrastructure. That pressure is not limited to Google. U.S. data centers consumed about 224 TWh in 2025, representing more than 5% of national electricity use. Meanwhile, North American peak demand is projected to rise by more than 224 GW over the coming decade, with new AI facilities contributing substantially to the increase. insight.tmcnet.com/insight/g…

City of Columbia officials are investigating a ransomware attack that disrupted public-facing municipal services, bringing in SLED and cybersecurity specialists to determine its scope. Officials have not disclosed which systems were compromised, how attackers obtained access, or whether sensitive information was removed from the city's environment. Those unanswered questions matter. During the opening stages of a ransomware response, investigators commonly need to preserve evidence, isolate affected infrastructure, identify compromised accounts, and distinguish encrypted systems from services taken offline as a precaution. Public disclosure can lag behind technical work, particularly when announcing details could interfere with containment or law enforcement activity. For residents and businesses, the immediate issue is service availability. Municipal technology environments can support billing, permitting, licensing, communications, records management, and other everyday functions. Even when core public-safety operations remain available, interruptions to administrative systems can create backlogs that persist after individual applications return. The available information does not identify which City of Columbia services were affected. The broader threat picture offers some context. Government ransomware incidents increased from 95 between April 2023 and April 2024 to 322 between April 2024 and April 2025, a 235.4% rise, according to Zscaler ThreatLabz data included in current public-sector threat research. Ransomware also appeared in 30% of the public-sector breaches covered by Verizon's 2025 dataset. Municipal agencies often operate a mixed estate of legacy applications, cloud services, employee endpoints, contractor access, and specialized departmental systems. That variety can make containment and restoration slower. It also raises a practical question: which services should return first when multiple departments depend on the same identity, network, or database infrastructure? insight.tmcnet.com/insight/c…

An Instagram post alleging a State Bank of India data breach is circulating against an already tense backdrop for digital banking security in India. So far, the available information does not establish that SBI systems were compromised or that customer data was disclosed without authorization. Unless State Bank of India or the Reserve Bank of India confirms an incident, the claim should be treated as unverified. That distinction matters. A social-media warning can refer to several very different situations: a direct intrusion into bank infrastructure, data stolen from a third-party provider, recycled credentials from an unrelated incident, or a phishing campaign merely using SBI’s name. Screenshots and supposed dark-web references can look persuasive while providing little verifiable information about the source, age or authenticity of the records. Even an unverified breach claim can create a genuine security risk. Fraudsters frequently exploit public anxiety by sending fake account-protection notices, password-reset messages or Know Your Customer update requests. A viral post can therefore become bait for a second-stage campaign, even when the original allegation lacks supporting evidence. Customers may be directed to counterfeit SBI pages or persuaded to disclose passwords, card details, one-time passcodes and UPI PINs. The broader fraud environment explains why the claim is attracting attention. According to Reserve Bank of India data for 2025-26, banks and financial institutions reported 10,114 fraud cases involving ₹48,021 crore. State-owned banks accounted for 5,418 cases. These totals cover multiple forms of fraud and do not amount to evidence of an SBI breach, but they illustrate the financial exposure confronting India’s banking sector. The number and value of incidents also tell different stories. MediaNama reported that card, internet and digital-payment frauds represented 66.8% of banking-fraud cases by number in 2024-25, comprising 7,756 incidents involving ₹252 crore. Separately, reported bank-fraud value increased 30% year over year to ₹21,515 crore during April through September 2025, despite a decline in case volume. insight.tmcnet.com/insight/u…

NextDC has moved to calm bondholders over its financial capacity, saying it has enough funding to get its data centre developments back on track as global capital markets grow more cautious about the sector's rapid expansion. The reassurance matters beyond NextDC. Data centre operators are being asked to build larger facilities, secure more electricity and deploy expensive computing infrastructure, often before all the expected revenue has materialised. Artificial intelligence demand has strengthened the commercial case for new capacity, but it has also increased the amount of capital required upfront. For bondholders, that combination creates a fairly direct concern: how much more funding will be needed before new developments begin generating stable cash flow? NextDC's response indicates that it believes its existing funding position is sufficient to support its development program. Still, jittery markets are applying greater scrutiny to construction schedules, liquidity reserves and the timing of customer demand. The scale of the global build-out helps explain the nerves. According to the International Energy Agency, five major technology companies spent more than $400 billion on capital expenditure in 2025, with a further 75% increase expected in 2026. The IEA argues that data centre investment has become too large to depend solely on corporate balance sheets. That is changing how infrastructure gets funded. Hyperscaler gross bond issuance exceeded $100 billion in 2025, while leases, capacity commitments, guarantees and special-purpose vehicles are taking a larger role. The Bank for International Settlements has noted that some of these off-balance-sheet arrangements can function economically like debt, even when their presentation differs from conventional borrowing. insight.tmcnet.com/insight/n…

Higgsfield has secured new investment from Nvidia and TCV, extending the AI video and image generation startup’s Series B financing as competition intensifies across generative media. The amount of the additional investment was not disclosed. The extension follows Higgsfield’s August 2026 Series B, which raised $400 million and valued the business at $5.4 billion. Variety reported that Nvidia joined TCV in the latest investment. The deal connects Higgsfield with the dominant supplier of computing infrastructure used to train and operate many generative AI systems, while bringing additional institutional backing to its expansion. Founded in 2023, Higgsfield has grown unusually quickly even by generative AI standards. The startup reports more than 40 million registered users and says its platform is approaching $1 billion in annualized revenue, according to Bloomberg. Annualized revenue is a run-rate measurement rather than recognized full-year revenue, but the figure still signals substantial demand for tools that can generate commercial video and imagery. Video generation is expensive. Producing consistent motion, characters and visual styles across sequences requires considerable computing capacity. For Higgsfield, Nvidia’s participation could carry strategic value beyond the undisclosed capital, although the companies have not announced specific infrastructure commitments or commercial terms. For Nvidia, the investment provides exposure to another fast-growing customer category for accelerated computing. It is not Nvidia’s only wager in the segment. Nvidia has also invested in Runway, which raised $315 million at a $5.3 billion valuation in February 2026. Runway has pursued professional entertainment applications, including an AI collaboration with Lionsgate. Backing multiple developers suggests Nvidia is positioning itself across the generative-video ecosystem rather than choosing a single likely winner. insight.tmcnet.com/insight/h…

The UK Cabinet Office has established the AI Biosecurity Lab, which it describes as the world’s first organisation dedicated specifically to artificial intelligence and biological security. According to the UK Cabinet Office, the lab is backed by £100 million in government funding and up to £200 million in commitments from OpenAI and Anthropic. Those company commitments include model credits supporting access to advanced AI systems, rather than being presented solely as direct cash investment. The lab sits within a broader National AI Biosecurity Initiative valued at £165 million. That total incorporates the government’s £100 million allocation for the lab, alongside £65 million for biosurveillance and medical countermeasures focused on priority pathogens. The distinction matters: the frequently cited figure of up to £300 million combines public funding with company commitments, while £165 million describes the wider government initiative. OpenAI and Anthropic are the initial named industry supporters. Google DeepMind and Basecamp Research have also been identified as prospective partners, giving the programme potential access to expertise spanning general-purpose models, biological foundation models and large-scale biological datasets. AI creates a two-sided biosecurity challenge. The same systems that help researchers analyse genomes, identify promising drug candidates or model outbreaks can also make complicated biological workflows easier to navigate. That does not mean current models can reliably produce pandemic-scale pathogens, as the available evidence is considerably more measured. insight.tmcnet.com/insight/u…