IT Week in Review: Agentic AI Ascends, Cyber Threats Persist, and Hardware Evolves (Week Ending May 23, 2025)

Introduction: A Week of Transformative AI, Persistent Cyber Threats, and Hardware Evolution

The week ending May 23, 2025, was marked by a relentless push in Artificial Intelligence, particularly the rise of “agentic AI” designed to act more autonomously. This rapid innovation, however, unfolded against a backdrop of persistent and sophisticated cybersecurity threats that challenged organisations globally. The hardware sector also saw significant unveilings, promising next-generation capabilities for consumers and enterprises alike. Broader market dynamics, influenced by economic forecasts and policy discussions, added another layer of complexity to the tech landscape. This period of intense technological advancement, intertwined with significant operational and ethical challenges, signals a maturing yet still volatile industry. Businesses and consumers alike must navigate a terrain rich with opportunity but also laden with risk, underscoring the need for informed agility. This report provides an accessible overview of these key developments, breaking down complex events into understandable insights.

The AI Revolution Accelerates: Agents, Devices, and New Frontiers

The Artificial Intelligence sector witnessed a flurry of significant announcements, underscoring a clear trajectory towards AI systems that are more autonomous, integrated, and specialized. From foundational platform shifts to new hardware paradigms and industry-specific applications, the week highlighted an accelerating revolution.

The Rise of Agentic AI: A Paradigm Shift in Computing

A dominant theme this week was the emergence and strategic focus on “agentic AI”—systems designed not just to respond, but to proactively plan and execute tasks to achieve user-defined goals. Microsoft is at the forefront of this shift, with CEO Satya Nadella articulating a vision where AI agents proliferate in 2025. The company is reportedly re-architecting its Windows operating system to be an agentic AI platform, a move aimed at building the technical foundation for an “open, agentic web”.¹ This represents one of the most significant overhauls in the operating system’s history. Furthering this vision, Microsoft also unveiled NLWeb, an open-source project designed to help developers create conversational AI interfaces for their websites using any model or data source, with NLWeb endpoints also functioning as Model Context Protocol (MCP) servers, making content easily discoverable by AI agents.²

Google is also making strides in this domain, unveiling plans to embed “agentic checkout” capabilities within its Search function. This development, announced at the Google I/O conference, aims to transform Search into a more integrated AI assistant, capable of browsing products, facilitating virtual try-ons, and even completing purchases with user confirmation or autonomously.¹ This move is part of Google’s broader strategy to evolve Search from an information retrieval tool to an intelligent assistant.

The broader industry sentiment, as captured in discussions and analyst reports, echoes this focus. Experts suggest that 2025 is shaping up to be the “year of agentic exploration,” with numerous companies, including Salesforce with its Agentforce platform, and hundreds of startups actively experimenting with and developing AI agents.³ Gartner has also identified “Agentic AI” as a top strategic technology trend for 2025, envisioning a virtual workforce of agents assisting, offloading, and augmenting human work.⁴

This concerted push by major technology players towards agentic AI indicates a fundamental evolution from AI as a passive tool to AI as an autonomous partner. Such a shift has the potential to redefine how users interact with technology and could automate a wide array of complex workflows across various sectors. The prospect of a “virtual workforce of agents” ⁴ could transform numerous job roles and industries. However, the path to truly autonomous and reliable agentic AI is not without significant hurdles. Experiments, such as the one conducted by Carnegie Mellon researchers where an AI-run virtual company saw agents excel at cerebral tasks but fail at simple, human-intuitive ones ¹, highlight current limitations. Experts emphasise that building AI agents capable of autonomously handling complex decision-making will require substantial advancements in contextual reasoning and comprehensive testing for edge cases.³ There’s also a degree of scepticism, with some questioning if “agentic AI” is fundamentally different from established concepts like “orchestration” and concerns persist about AI’s ability to correctly interpret nuanced human communication.³ The return on investment for Large Language Model (LLM) technology more broadly is also yet to be fully established, adding a layer of financial caution to the ambitious technological visions. Gartner underscores the necessity of “robust guardrails” to ensure these autonomous systems align with user and provider intentions, acknowledging the “terrifying” prospect of AI making critical decisions without adequate oversight.³ The successful realisation of an “agentic web” may therefore depend heavily on addressing these fundamental issues of reliability, control, and human-AI interaction.

New AI Models and Platforms: Expanding Capabilities

The development of new and more powerful AI models continued at a rapid pace. Anthropic launched Claude Opus 4 and Claude Sonnet 4, models capable of undertaking long-running tasks and working continuously for several hours. Claude Opus 4 is noted for its prowess in coding and complex problem-solving, while Sonnet 4 improves upon its predecessor in balancing performance and efficiency. Alongside these models, Anthropic announced a beta for extended thinking with tool use, parallel tool use capabilities, and the general availability of Claude Code.² The Anthropic API also received enhancements, including a code execution tool and a Files API.²

Google’s I/O conference was a showcase for its latest AI innovations. The company introduced new models such as Gemini Diffusion, a text model, and Gemma 3n, a multimodal model designed for on-device operation (phones, laptops, tablets) capable of processing audio, text, image, and video. Google also revealed specialised Gemma model variants: MedGemma for healthcare applications and SignGemma for translating sign language into spoken language text.² Furthermore, Gemini Code Assist for individuals and GitHub, powered by Gemini 2.5, became generally available, offering features like chat history, custom commands, and partial code acceptance.²

Microsoft is bolstering the AI development ecosystem with its Windows AI Foundry, a platform designed to support the entire AI developer lifecycle, from training to inference. This will allow developers to manage open-source LLMs locally or bring proprietary models for fine-tuning and deployment across clients and the cloud.² Microsoft also added support for the Model Context Protocol (MCP) across its platforms, including GitHub, Copilot Studio, and Windows 11, to enhance interoperability.²

In the realm of developer tools, GitHub Copilot introduced a new coding agent. This agent can be activated by assigning it a GitHub issue or through a prompt in VS Code, assisting with tasks such as adding features, fixing bugs, extending tests, refactoring code, and improving documentation. Importantly, all pull requests generated by this agent require human approval before execution.²

The proliferation of new and increasingly specialised AI models, such as multimodal systems and those tailored for specific industries like healthcare or accessibility needs like sign language translation, marks a significant trend. This indicates a maturation in AI development, moving beyond general-purpose LLMs towards solutions designed for specific use cases. This specialisation is likely to accelerate AI adoption in various niche sectors. However, it also brings to the forefront the critical need for rigorous validation and careful ethical consideration, particularly when deploying AI in sensitive domains such as medicine.

AI Hardware and Infrastructure: Powering the Revolution

The insatiable demands of AI are profoundly influencing hardware development and infrastructure strategies. In a landmark move, OpenAI is reportedly acquiring Io, an AI devices startup founded by former Apple Chief Design Officer Jony Ive and other Apple designers, for a reported $6.4 billion in equity. Io is expected to become OpenAI’s hardware division, with the goal of reinventing devices for the AI era. OpenAI CEO Sam Altman has lamented the current cumbersome process of accessing tools like ChatGPT and believes a more integrated hardware approach is necessary.¹

The chip architecture landscape is also shifting. Arm is projecting that it will capture 40% of PCs and tablets shipped this year. Its power-efficient architecture, already dominant in smartphones, is increasingly seen as crucial for next-generation AI-enabled devices, particularly due to the advantages of edge computing which enhances privacy, reduces latency, and cuts cloud costs.¹

The immense energy consumption of AI operations is another critical consideration. Meta (parent company of Facebook, Instagram, and WhatsApp) announced a substantial investment in renewable energy, securing a deal to buy 650 megawatts (MW) of solar power from AES to power its expanding AI and data centres in Texas and Kansas.⁵ This is part of Meta’s broader commitment to achieve net-zero emissions across its value chain by 2030 and adds to its existing portfolio of over 12 gigawatts (GW) of clean energy capacity.⁵ The scale of AI’s energy needs was further underscored by the U.S. Energy Secretary, Chris Wright, who emphasised the potential of nuclear power to meet the skyrocketing electricity demand from data centres powering AI.⁶

The AI boom is thus creating a symbiotic relationship: a demand for novel AI-centric hardware, exemplified by the OpenAI/Jony Ive venture, and an equally intense demand for massive, sustainable energy solutions to power the underlying infrastructure. The future growth of AI is inextricably linked to breakthroughs in both energy production and hardware efficiency. Failure to address the energy demand sustainably could become a significant bottleneck for AI development or lead to severe environmental repercussions. Concurrently, the drive for AI-native devices and the increasing market share of Arm in the PC sector suggest a potential transformation in the personal computing landscape. AI capabilities are becoming a primary driver of hardware design and consumer choice, potentially challenging long-standing dominances, such as Intel’s in the PC market. A new wave of hardware innovation focused on AI-specific functionalities could emerge, altering competitive dynamics among chip manufacturers and device makers.

AI in Action: Industry Applications and New Alliances

AI is rapidly transitioning from theoretical potential to practical application across a multitude of industries. A new alliance named “Embracing AI,” formed by technology leaders, officially launched with the mission to simplify and accelerate operational AI adoption for organisations of all sizes, helping them turn data into decision-ready insights securely and reliably.⁷

In the media sector, ViewLift unveiled two AI-powered solutions designed to revolutionise business intelligence and customer support operations for media companies and content owners.⁷ The pharmaceutical industry is also seeing AI-driven innovation, with Persist AI raising $12 million in Series A funding for its Cloud Lab. This platform allows pharmaceutical companies to develop formulations using AI and robotic laboratory facilities via a website, aiming to dramatically accelerate drug development.⁷

Manufacturing and supply chain management are being transformed as well. Palantir and Divergent formed a partnership to integrate Divergent’s Adaptive Production System (DAPS™) with Palantir’s software. This collaboration aims to enable customers to rapidly identify emerging supply chain vulnerabilities and address them through on-demand manufacturing.⁷

The software development field is also benefiting from AI. CodeBuddy, a Nebraska-based startup, secured $872,500 in seed funding to expand its AI-powered software development platform. This platform is designed to assist nontraditional developers in creating new software from concept to cloud, blending AI tools with expert mentorship.⁸

These examples demonstrate AI’s move into operational integration across diverse sectors like media, pharmaceuticals, manufacturing, and software development. A notable trend is the emergence of platforms and alliances that facilitate broader AI adoption, particularly for small and medium-sized businesses (SMBs). This widespread adoption promises to drive efficiency and innovation. However, it also brings challenges, including the need to upskill the workforce to effectively utilise these new tools ⁹ and to navigate the complex industry-specific regulatory and ethical considerations that arise with AI implementation.

AI Challenges: Performance, Risks, and Ethical Considerations

Despite the rapid advancements, significant challenges remain in AI development and deployment. The aforementioned Carnegie Mellon study, where AI agents in a virtual software company struggled with simple, human-intuitive tasks—with Anthropic’s Claude, the best performer, completing only 24% of assigned tasks—underscores a gap between AI’s current capabilities and the ambition for fully autonomous systems.¹

The reliability of AI outputs is a major concern. A Liberty Mutual executive warned of “Hallucination Liability” as a new corporate risk associated with AI adoption, referring to instances where AI models generate incorrect or fabricated information.¹⁰ This highlights the potential for significant errors if AI systems are not properly validated and overseen.

The very infrastructure underpinning the AI boom is also facing scrutiny. An article in The New Stack, titled “Cloud Realities Are Slowing AI Ambitions,” argues that current cloud computing setups are not optimally designed to meet the unique and evolving performance requirements of AI workloads, particularly the highly interactive and latency-sensitive nature of Agentic AI.¹¹ This suggests that the cloud infrastructure which enabled the current AI surge might become a bottleneck for next-generation AI if not re-optimised.

Broader risks and ethical considerations are also gaining prominence. Analyst reports from Gartner and Info-Tech emphasise the growing need for robust AI governance frameworks, strategies for deepfake defence, and proactive engagement with emerging AI regulations.⁴ These concerns point to the necessity of developing AI responsibly, with safeguards against misuse and unintended consequences.

The gap between AI’s current abilities and the aspiration for fully autonomous, dependable systems is evident. While AI shows promise in complex analytical tasks, its limitations in common-sense reasoning and consistency can lead to operational failures, financial liabilities, and an erosion of public trust if not managed carefully. Over-reliance on immature AI systems without robust testing, validation, and human oversight remains a critical risk. Furthermore, the potential for existing cloud infrastructure to hinder the progress of advanced AI applications like Agentic AI suggests a need for a new wave of cloud innovation. This could involve developing AI-specific hardware, ultra-low-latency networking solutions, and more sophisticated edge computing capabilities, potentially reshaping the competitive landscape among cloud providers and requiring organisations to rethink their cloud strategies for AI.

Key AI Announcements of the Week

Cybersecurity Frontline: Threats, Defences, and Global Preparedness

The cybersecurity landscape remained highly active, with reports of significant breaches, ongoing campaigns by sophisticated threat actors, and new defensive measures being developed and tested.

Major Incidents and Exploited Vulnerabilities

A particularly concerning development involved Chinese-speaking hackers, tracked as UAT-6382, exploiting a zero-day vulnerability (CVE-2025-0994, CVSS score 8.6) in Trimble’s Cityworks asset management software. This software is widely used by local governments and utilities organisations across the United States to manage infrastructure and community services.¹² The attackers exploited this flaw, patched in late January, to remotely execute code and gain administrative access to enterprise networks, deploying malware such as various web shells (AntSword, Chinatso, Behinder) and tools like Cobalt Strike and a GoLang-based implant called VShell to maintain long-term access and conduct reconnaissance.¹² Reports indicate a clear interest by the attackers in pivoting to systems related to utilities management.¹² The Cybersecurity and Infrastructure Security Agency (CISA) had previously issued an advisory about this vulnerability in February.¹² This incident underscores the significant national security risk posed by the targeting of critical infrastructure, often through under-defended entry points in local government IT systems.¹⁴ The use of a zero-day vulnerability points to the advanced capabilities and resources of the attackers, likely state-sponsored, aiming for espionage or potential disruption of essential services.

In other incidents, Cellcom, a Wisconsin-based mobile carrier, confirmed that a cyberattack was the cause of widespread voice and SMS service outages that began on May 14, affecting customers across Wisconsin and Upper Michigan. The company stated that sensitive customer data was not impacted and that it was working with the FBI and cybersecurity experts to restore services.¹⁵

A different kind of threat emerged from the hardware supply chain. Malware, including a worm, the Floxif file infector, a backdoor, and a crypto-stealing Trojan named SnipVex, was discovered in official software downloads and on USB drives included with UV inkjet printers from the brand Procolored. Cybersecurity firm G Data reportedly traced approximately $100,000 in stolen Bitcoin linked to the SnipVex Trojan. Procolored later acknowledged that malware might have been introduced via USB and has since cleaned its downloads.¹⁵ This incident highlights that supply chain security must extend beyond software to physical hardware components and their bundled accessories, as even seemingly innocuous peripherals can introduce significant risks. It implies a need for organisations to scrutinise all hardware and software, regardless of source, and for manufacturers to tighten security across their own operations and supplier networks.

Additionally, a major global outage affecting an estimated 8 million Microsoft Windows-based computers was attributed to a defective update to CrowdStrike’s Falcon cybersecurity software. While not a malicious cyberattack in the traditional sense, this event had significant operational security implications. Delta Air Lines, for example, estimated $550 million in losses due to the outage, with only $50 million recouped through fuel savings.¹⁵ This highlights the critical reliance on cybersecurity software and the potential for widespread disruption if such software fails or is improperly updated.

Emerging Threats and Attack Vectors

Beyond specific incidents, new and evolving threat actor campaigns were observed. The SideWinder Advanced Persistent Threat (APT) group launched a targeted cyber-espionage campaign against government institutions in Sri Lanka, Bangladesh, and Pakistan. This campaign utilised spear-phishing emails and geofenced malware payloads, exploiting outdated Microsoft Office vulnerabilities to deliver the StealerBot malware, a.NET-based tool designed to capture sensitive data.¹⁵

A China-linked threat actor dubbed “UnsolicitedBooker” was found to have used a novel backdoor named “MarsSnake” in a multi-year espionage campaign targeting an unnamed international organisation in Saudi Arabia. The MarsSnake backdoor, which allows for persistent access and data exfiltration, reportedly remained undetected for several years.¹⁵

The security of mobile ecosystems also came under scrutiny. A new report from Zimperium warned that iOS devices, often perceived as highly secure, are increasingly being targeted through sideloaded and unvetted applications. Attackers are exploiting flaws in iOS using tools like TrollStore and SeaShell, and vulnerabilities such as MacDirtyCow and KFD, to bypass Apple’s built-in protections. These malicious apps, which may appear benign, can exfiltrate data or compromise devices without detection. Zimperium found over 40,000 apps using private entitlements and more than 800 using private APIs, posing serious risks.¹⁵

Furthermore, hackers are employing new fileless techniques to deploy the Remcos Remote Access Trojan (RAT) malware. This method involves a PowerShell-based loader and is designed to bypass Windows Defender, starting with a malicious ZIP file containing a spoofed LNK shortcut.¹⁵

The continued refinement of tactics by threat actors, leveraging both old vulnerabilities (like unpatched Office flaws) and novel methods (new backdoors, fileless techniques, iOS sideloading exploits), demonstrates a persistent and adaptive adversary landscape. This diversity in attack vectors means that organisations cannot afford to be complacent. Effective cybersecurity requires a multi-layered defence strategy that includes robust patch management for known vulnerabilities, advanced threat detection capabilities to identify novel attacks, and comprehensive user education, especially concerning emerging risks like those associated with sideloaded applications on mobile platforms. The perception of certain platforms as inherently “safe” can create a dangerous false sense of security.

Industry Responses and Defensive Measures

In response to the evolving threat landscape, the cybersecurity industry and international bodies are actively developing more predictive and collaborative defence mechanisms. Resecurity, a global cybersecurity leader, announced a strategic partnership with Starlink to expand advanced threat intelligence and cybersecurity solutions across the Middle East and Africa, aiming to enhance regional cybersecurity capabilities.¹⁵

Researchers from CISA and NIST introduced a new metric called “Likely Exploited Vulnerabilities” (LEV). This metric aims to improve patch prioritisation by better predicting which software flaws are being actively exploited, combining data from various sources like the Exploit Prediction Scoring System (EPSS) and Known Exploited Vulnerabilities (KEV) lists.¹⁵

On the product front, Superna launched Data Security Essentials for Dell PowerStore. This solution is designed to provide real-time visibility and control to reduce risk from ransomware attacks before they begin and to stop threats as they strike.⁷

International cooperation and preparedness were highlighted by NATO’s Locked Shields 2025, the world’s largest live-fire cyber defence exercise. Held in Tallinn, Estonia, the exercise involved approximately 4,000 experts from 41 countries participating remotely. They simulated the defence of over 8,000 systems against thousands of cyberattacks, with scenarios including AI-driven narratives and quantum computing challenges.¹⁵ Such initiatives indicate a shift from purely reactive defence postures to more proactive and collective security approaches, recognising the global nature of cyber threats and the need for shared intelligence and coordinated responses. The inclusion of AI and quantum computing scenarios in exercises like Locked Shields demonstrates foresight in preparing for future, more sophisticated threat vectors.

The growing emphasis on “AI-powered defences,” as noted by Info-Tech Research Group ⁹, and the development of tools like Superna’s offering suggest that AI is becoming a critical component on both sides of the cybersecurity equation. As AI tools become more accessible, they will inevitably be used to automate and enhance both offensive and defensive cyber operations. This creates an “AI arms race” in cybersecurity, where defenders must continually innovate to stay ahead of AI-driven attacks and cybersecurity professionals will increasingly require AI-related skills.

Notable Cybersecurity Incidents & Developments

Hardware Highlights: Innovations from Computex and Beyond

The hardware sector buzzed with activity, largely driven by announcements from Computex 2025 in Taipei, alongside other notable developments in consumer electronics and enterprise hardware.

Computex 2025: A Glimpse into Next-Gen Tech

Computex served as a platform for major brands to unveil their latest innovations, with a clear focus on powering AI and high-end gaming experiences, alongside a growing emphasis on aesthetics and user customization.

• AMD’s Showcase: Advanced Micro Devices (AMD) made a significant splash with the introduction of its new Ryzen Threadripper 9000 Series processors. The flagship model, the Ryzen 9 9995WX, boasts an impressive 96 cores and 192 threads, with peak boost clocks reaching 5.4GHz and a 300W TDP, targeting high-end desktop (HEDT) and workstation users with demanding computational needs.¹⁶ On the graphics front, AMD launched the new Radeon RX 9060 XT GPU, built on the Navi 44 architecture. Starting at $299 for the 8GB version and $349 for the 16GB model, and featuring a boost clock of 3,130 MHz, this GPU is positioned as a strong contender for the value segment in 2025.¹⁶ AMD also discussed FSR Redstone, a major update to its FidelityFX Super Resolution technology, and unveiled the Radeon AI Pro R9700, a PCIe 5.0 GPU for workstations.¹⁶
• Cutting-Edge Monitors: The display market saw advancements aimed at gamers seeking ultra-smooth visuals. MSI showcased its X50 Gaming Monitor, a QD-OLED panel featuring a 500Hz refresh rate at 1440p resolution. A novel feature is an AI sensor that can power down the panel when it detects the user has walked away, helping to prevent burn-in.¹⁶ Asus presented the ROG Strix Ace XG248QSG, a 24-inch 1080p gaming monitor that can be overclocked to an extraordinary 610 Hz refresh rate using a “Super TN” panel.¹⁶
• Power Supply Units (PSUs): A clear trend towards ultra-high wattage PSUs was evident, driven by the power demands of next-generation CPUs and multiple GPU setups common in AI rigs and high-end gaming PCs. Both Asus and Super Flower displayed 3,000W power supplies.¹⁶ Lian Li introduced new RS-series PSUs with innovative rotatable power outlets for enhanced mounting flexibility ¹⁶, while Seasonic unveiled new Prime Power Supplies equipped with sensors to detect voltage issues and automatically shut down to prevent damage.¹⁶
• Innovative PC Cases: PC case design saw a blend of performance-focused engineering and aesthetic creativity. SilverStone garnered attention with its FLP02, a retro-style tower case evoking late-1980s designs, complete with faux floppy drives on the front, yet thoroughly modern internally.¹⁶ Corsair launched the Air 5400, featuring a triple-chamber layout designed for high-performance cooling by directly exhausting CPU heat via a 360mm AIO radiator.¹⁶ Other notable designs included Lian Li’s V200 case with glass panels and an embedded screen, and Thermaltake’s MineCube 360 AIO cooler with a customizable screen capable of displaying Minecraft scenes.¹⁶ Antec also presented its Vortex View AIO cooler with a 5-inch display on a ball joint for flexible positioning.¹⁶
• Other Noteworthy Products: V-Color introduced its Xfinity Manta DDR5 RAM modules featuring a full-colour LCD screen that can display system statistics like voltages and clock speeds.¹⁶ Asus showcased a creative take on the upcoming RTX 5080 with its ProArt version, accented with wood and offering an M.2 slot option for an SSD.¹⁶ Crucial announced its T710 SSD, aiming to be the fastest consumer SSD with significant read/write speed improvements.¹⁶ Intel also unveiled its new Arc Pro B50 and B60 workstation GPUs.¹⁶

The hardware innovations unveiled at Computex clearly reflect the immense processing power required by AI applications—evidenced by AMD’s 96-core Threadripper and the emergence of 3000W PSUs designed to handle multiple power-hungry GPUs. Similarly, the demands of high-end gaming continue to push boundaries in display technology with ultra-high refresh rate monitors and advanced cooling solutions. While these cutting-edge technologies often debut at the high end, they tend to trickle down to more mainstream consumer products over time. However, this rapid advancement also widens the gap between enthusiast-grade/professional hardware and standard components, and raises concerns about the escalating energy consumption of such high-performance systems, linking back to the broader discussions on AI’s energy footprint.

Beyond raw performance, a discernible trend in PC hardware is the growing emphasis on enhanced aesthetics, customisation, and even nostalgia. Products like SilverStone’s retro case, V-Color’s RAM with integrated LCDs, and Asus’s wood-accented GPU, alongside cases featuring unique visual designs and embedded screens, suggest that user experience and personalisation are becoming increasingly important differentiators. As performance across similar-tiered components becomes more comparable, manufacturers are seeking alternative ways to attract consumers by appealing to their sense of style, individuality, or a fondness for classic designs. This indicates an evolution of the PC from a purely functional tool to a more personalised form of expression for many enthusiasts, likely leading to more diverse and visually interesting product designs, albeit potentially at a premium price point.

Other Notable Hardware Developments

Outside of Computex, Samsung announced that its 2025 OLED TV lineup will feature NVIDIA G-Sync compatibility, a significant enhancement for clients with gaming rigs, allowing the TV’s refresh rate to synchronise with the GPU’s frame rate for smoother gameplay. Samsung is also adding AMD’s equivalent, FreeSync Premium Pro, to its lineup.¹⁷

In the defence sector, Solute Inc., based in San Diego, California, was awarded a $41,382,793 indefinite-delivery/indefinite-quantity contract by the U.S. Navy. This contract is for providing engineering support for Consolidated Afloat Networks and Enterprise Services (CANES), encompassing technical and programmatic services for networking.¹⁸

In corporate news, shareholders of SatixFy, an Israeli company specialising in satellite communication systems, overwhelmingly approved its proposed acquisition by MDA Space, a Canadian space technology company.¹⁹ Meanwhile, ONERugged announced it would be exhibiting its rugged AI and industrial terminals at COMPUTEX 2025, showcasing hardware designed for demanding environments and digital transformation in key industries.¹⁹

These developments in specialised hardware for sectors like defence and industrial AI, coupled with enhancements in consumer electronics such as improved gaming experiences on televisions, demonstrate the pervasive reach of hardware innovation. The IT hardware industry is not a monolith but comprises diverse segments, each with unique requirements and innovation cycles, ranging from large-scale enterprise and government contracts to consumer-facing improvements.

Selected Hardware Launches from Computex 2025

Cloud, Enterprise Software, and Big Tech Moves

The cloud computing and enterprise software sectors saw significant strategic maneuvering, including potential major acquisitions and ongoing adaptation to the demands of AI. Big Tech companies also made headlines with substantial investments and market-impacting news.

Cloud Dynamics and Enterprise Software

A major story in the enterprise software space is the renewed interest from Salesforce in acquiring Informatica, a provider of data management and integration tools. Talks for an $11 billion deal reportedly fell apart last year, but Salesforce is said to be pursuing Informatica again. This time, however, Salesforce may face competition from Cloud Software Group (CSG), the parent company of Citrix. According to reports, a deal with Salesforce could be announced as soon as next week.²⁰ Informatica has demonstrated strong growth in its cloud offerings, with its cloud subscription annualised recurring revenue (ARR) increasing by 30% year-over-year to $848 million in its latest quarterly report.²⁰ This potential acquisition underscores the increasing strategic importance of robust data management and integration capabilities, especially as AI technologies, which are heavily reliant on high-quality, well-governed data, become more central to enterprise operations. Salesforce’s concurrent plan to acquire Convergence.ai, a London-based AI agent startup, to bolster its Agentforce platform further emphasises this data-AI synergy.²⁰ An acquisition of Informatica would significantly strengthen Salesforce’s ability to offer end-to-end, data-driven AI solutions. This signals a broader consolidation trend where large enterprise software vendors aim to control more of the data and AI stack to provide comprehensive and powerful solutions.

The critical role of cloud infrastructure in supporting AI was a recurring theme. As highlighted previously, the article “Cloud Realities Are Slowing AI Ambitions” argues that current cloud setups, optimised for traditional digital applications, require re-optimisation to meet the unique performance demands of AI, particularly the latency-sensitive nature of Agentic AI.¹¹ This suggests that the cloud layer is actively evolving. Red Hat, for instance, announced that its AI Platform now includes an AI Inference Server, and its upcoming Red Hat Enterprise Linux 10 is being positioned as an AI-driven and quantum-ready platform.¹¹ These developments indicate a shift towards more specialised cloud offerings designed to provide the necessary performance, scalability, and specific services required by burgeoning AI workloads, moving beyond generic compute and storage. This will likely intensify competition among cloud providers to offer the best AI platforms and necessitate careful architectural considerations for organisations embarking on AI initiatives.

Big Tech Developments and Market Impacts

Meta continued to make headlines with its strategic investments in infrastructure to support its AI ambitions. The company’s commitment to purchase 650 MW of solar power from AES for its data centres in Texas and Kansas is a significant move towards its goal of achieving net-zero emissions by 2030 and powering its operations with 100% renewable energy.⁵ Meta already claims over 12 GW of clean energy capacity through various solar and wind projects.⁵ Such substantial, long-term investments reflect both the technological imperative of powering energy-intensive AI operations and broader corporate responsibility objectives around sustainability. This occurs even as these tech giants navigate short-term market volatility.

The stock market experienced turbulence this week, partly due to renewed tariff threats from former U.S. President Donald Trump. His comments about potential tariffs on EU imports and a specific threat that Apple would face “at least 25%” tariffs if iPhones sold in the U.S. are not manufactured domestically sent ripples through the market.⁶ Consequently, major tech stocks, including Apple (whose shares dropped 3% on the news, marking an eighth consecutive day of decline), Microsoft, Nvidia, Amazon, Alphabet, and Meta, all saw declines. The broader market indexes—the Dow Jones Industrial Average, S&P 500, and Nasdaq Composite—also posted losses for the week.⁶

Company earnings reports also influenced stock performance, illustrating that even within the tech sector, outcomes can diverge based on company-specific fundamentals and exposure to geopolitical factors. Shares of Intuit, the maker of TurboTax, jumped 8% following strong earnings and a positive outlook. The company could also potentially benefit from the proposed elimination of the free IRS direct tax filing system.⁶ Conversely, shares of Workday, a cloud-based software provider, sank approximately 13% due to a lacklustre subscription revenue outlook and concerns about stiffening competition in human resources and financial management software.⁶ Other companies like discount retailer Ross Stores and shoe brand parent Deckers Outdoor also saw their shares fall, citing uncertainties related to tariffs and their manufacturing operations in China.⁶

In contrast to the general downward trend for some tech stocks, cybersecurity firm CrowdStrike saw its shares advance 2.6% to reach a record-high close. This was attributed to positive recognition for its identity theft protection capabilities and an announcement of its integration into enterprise AI ecosystems powered by Nvidia’s Blackwell infrastructure.⁶ Meanwhile, Wedbush analysts expressed strong optimism for Tesla, raising their price target on the stock to $500, one of the highest on Wall Street, ahead of the electric vehicle maker’s planned June launch of fully autonomous Teslas in Austin, Texas.⁶

The varied market reactions demonstrate that while broad sentiment, such as tariff fears, can have a widespread impact, individual company performance, competitive positioning, and specific risk exposures (like supply chain concentration in China) remain crucial determinants of market valuation. This underscores the need for investors and analysts to look beyond sweeping sector trends and delve into the nuanced specifics of each company’s strategy, execution, and vulnerabilities.

Policy & Connectivity: Bridging the Digital Divide

Efforts to improve digital access and affordability were a focus of policy discussions this week. In Connecticut, the State Senate passed Senate Bill 514 by a significant majority (29 to 7). The bill aims to create more affordable internet access for low-income residents by encouraging broadband providers to offer a $40 per month plan, inclusive of all taxes and fees. Eligibility for this plan would be extended to residents enrolled in the Supplemental Nutrition Assistance Program (SNAP).²¹

To incentivise participation from broadband providers, the amended bill stipulates that those offering such an affordable option would be given preference in state contracts. Senator James Maroney, a proponent of the bill, described this as a “more stick than carrot approach”.²¹ During the Senate discussion, concerns were raised by Senator Paul Cicarella about whether the bill might lead to increased internet rates for users who do not qualify for the subsidised plan. However, Senator Maroney expressed his belief that this would not be the case, as most companies already provide some form of low-cost service option.²¹ The bill now moves to the House for further consideration.

Legislative efforts like Connecticut’s Bill 514 reflect a growing governmental recognition of affordable internet access not just as a convenience, but as a crucial utility and a matter of social equity. By leveraging the power of state contracting, the bill seeks to incentivise private sector involvement in bridging the digital divide. Such policies aim to promote digital inclusion, which is essential for education, employment opportunities, and access to vital services in an increasingly online world. However, these initiatives often face challenges regarding long-term sustainability, the extent of provider adoption, and potential unintended consequences on broader market pricing structures, despite assurances from lawmakers. The debate surrounding potential rate increases for other consumers is a common and valid concern associated with targeted subsidy programs.

Looking Ahead: Key IT Trends Shaping 2025

Analyst firms provided insights into the broader technological currents expected to define 2025 and beyond, painting a picture of transformative change driven by AI, quantum computing, and evolving human-machine interactions.

Gartner’s “Top Strategic Technology Trends for 2025” report organises its predictions across three core themes: AI imperatives and risks, new frontiers of computing, and human-machine synergy. Among the ten key trends identified are Agentic AI, which aligns with the week’s major announcements; Post-Quantum Cryptography (PQC), addressing the security risks posed by future quantum computers; Spatial Computing, which blends digital and physical realities; and AI Governance Platforms, highlighting the need for frameworks to manage AI responsibly. Other notable trends include Ambient Invisible Intelligence (technology seamlessly embedded in environments), Polyfunctional Robots, Disinformation Security, Energy-Efficient Computing (a critical concern given AI’s demands), Neurological Enhancement (such as brain-computer interfaces), and Hybrid Computing.⁴

Info-Tech Research Group’s “Tech Trends 2025” report also emphasises the pervasive influence of AI. It notes that Generative AI continues to dominate IT investment strategies, with 87% of organisations incorporating it into their plans to revolutionise areas like customer service, software development, and marketing. As GenAI matures, the report stresses the importance of responsible AI governance. Quantum computing is also gaining momentum, transitioning from theory to real-world application, with 33% of organisations in media, telecom, and technology sectors reportedly investing in it. This rise is accompanied by the critical need for post-quantum cryptography to safeguard sensitive data.⁹

A significant challenge highlighted by Info-Tech is workforce evolution. With 42% of IT professionals exploring new job opportunities, driven by evolving skill demands and the rise of generative AI, organisations face an urgent need for upskilling and reskilling employees to bridge experience gaps and foster innovation. The proliferation of deepfake technology and increasing regulatory focus on AI, such as the EU AI Act, present further challenges, requiring robust verification systems and compliance preparedness. Other key trends include the strategic alignment of IT initiatives with overall business goals, the rise of localised AI models for improved cost efficiency and control, and the evolution of cybersecurity towards AI-powered defences to combat increasingly complex threats.⁹

The convergence of insights from these analyst firms points to AI, in its various forms (Generative, Agentic, AI for governance, AI-powered cybersecurity), and Quantum Computing (along with the necessary defensive posture of PQC) as dominant and transformative forces for 2025. These are not isolated advancements but interconnected drivers of profound technological shifts. AI’s growth fuels the demand for greater computational power, which quantum computing may eventually provide, while quantum capabilities simultaneously introduce new security paradigms. This necessitates significant organisational adaptation, strategic investment in new technologies and skills, and proactive risk management. The challenges associated with these trends, such as the complexities of migrating to PQC, the quest for energy-efficient computing, and the development of robust AI governance, are substantial.

Furthermore, emerging trends like “Ambient Invisible Intelligence,” “Neurological Enhancement,” and “Polyfunctional Robots,” coupled with the overarching theme of “human-machine synergy” ⁴, suggest a future where technology becomes even more deeply and seamlessly integrated into human lives and work environments. This deeper integration blurs the lines between the physical, digital, and even biological realms. While offering potentially transformative benefits in efficiency, personalisation, and human capability augmentation, these trends will inevitably raise complex ethical, societal, privacy, and security questions, particularly concerning technologies like neurological interfaces that interact directly with the human brain. The “ethical concerns” and “data privacy” challenges flagged by Gartner will become increasingly paramount as these frontiers are explored.

Conclusion: Navigating a Dynamic IT Landscape

The IT industry, as evidenced by the developments of the week ending May 23, 2025, is in a state of dynamic flux, characterised by an accelerating pace of innovation, particularly in Artificial Intelligence. AI is acting as a central catalyst, driving profound changes across hardware design, software development, cloud infrastructure requirements, and even global energy demands. The rise of agentic AI, new powerful models, and AI-specific hardware initiatives dominated the headlines, promising a future of more autonomous and integrated intelligent systems.

However, this rapid progress is inextricably linked with escalating complexities. Cybersecurity remains a critical battleground, with sophisticated threat actors exploiting new vulnerabilities and targeting critical infrastructure, even as the industry develops more advanced and collaborative defence mechanisms. The hardware sector is racing to meet the voracious demands of AI and high-end computing, leading to powerful new processors, graphics cards, and innovative system designs, but also raising concerns about energy consumption.

The week also highlighted the significant influence of economic conditions and policy decisions on the tech sector. Market volatility in response to potential trade policy shifts and the varied performance of tech companies based on their individual earnings and geopolitical exposure underscore the intricate relationship between technology and broader global factors. Furthermore, policy initiatives aimed at bridging the digital divide, such as efforts to ensure affordable internet access, reflect a growing societal expectation that the benefits of technology should be widely accessible.

The interconnectedness of these developments is a key takeaway. Advances in AI fuel the need for more powerful and energy-efficient hardware and specialised cloud services. This increased reliance on complex digital infrastructure, in turn, creates new attack surfaces and necessitates more sophisticated cybersecurity measures. The ethical implications of increasingly autonomous AI and human-machine integration are also becoming more pronounced, demanding careful consideration and robust governance.

Ultimately, the IT landscape in May 2025 presents both immense opportunities and considerable challenges. Navigating this environment successfully requires businesses, policymakers, and individuals to remain informed, adaptable, and strategically forward-thinking. The ability to understand these complex interdependencies and to anticipate the trajectory of key trends will be crucial for harnessing the transformative potential of technology while mitigating its inherent risks.

Disclaimer

This article is a summary of news and developments in the Information Technology industry for the week ending May 23, 2025, based on a selection of publicly available research materials. It is intended for informational purposes only and should not be construed as financial, investment, legal, or professional advice of any kind.

While efforts have been made to ensure the accuracy and completeness of the information presented, based on the source materials provided, this publication assumes no responsibility or liability for any errors or omissions in the content of this article or the source materials themselves. The information is provided on an “as is” basis with no guarantees of completeness, accuracy, usefulness, or timeliness.

Readers are advised to consult the original sources of information and to seek professional advice for any specific concerns or decisions. Reliance on any information provided in this article is solely at the reader’s own risk. This publication is not responsible for any actions taken, or not taken, based on the content of this report, nor for any results obtained from the use of this information.

References

1. This Week in AI: The Agentic Web and Search Tools – PYMNTS.com, accessed on May 24, 2025, https://www.pymnts.com/artificial-intelligence-2/2025/this-week-in-ai-the-agentic-web-ai-native-devices-and-search-tools/
2. AI updates from the past week: Anthropic launches Claude 4 models …, accessed on May 24, 2025, https://sdtimes.com/ai/ai-updates-from-the-past-week-anthropic-launches-claude-4-models-openai-adds-new-tools-to-responses-api-and-more-may-23-2025/
3. AI Agents in 2025: Expectations vs. Reality | IBM, accessed on May 24, 2025, https://www.ibm.com/think/insights/ai-agents-2025-expectations-vs-reality
4. Explore Gartner’s Top 10 Strategic Technology Trends for 2025, accessed on May 24, 2025, https://www.gartner.com/en/articles/top-technology-trends-2025
5. Meta Invests in 650 MW of Solar Energy to Power AI and Data Centers, accessed on May 24, 2025, https://carboncredits.com/meta-invests-in-650-mw-of-solar-energy-to-power-ai-and-data-centers/
6. Markets News, May 23, 2025: Stocks Fall as Latest Trump Tariff …, accessed on May 24, 2025, https://www.investopedia.com/dow-jones-today-05232025-11740754
7. Weekly Recap: 15 Tech Press Releases You Need to See, accessed on May 24, 2025, https://www.prnewswire.com/news-releases/weekly-recap-15-tech-press-releases-you-need-to-see-302463765.html
8. CodeBuddy closes $872k seed round to expand AI-powered software development platform, accessed on May 24, 2025, https://siliconprairienews.com/2025/05/codebuddy-closes-872k-seed-round-to-expand-ai-powered-software-development-platform/
9. Top Technology Trends to Watch in 2025: Insights from Info-Tech Research Group, accessed on May 24, 2025, https://www.prnewswire.com/news-releases/top-technology-trends-to-watch-in-2025-insights-from-info-tech-research-group-302349501.html
10. Liberty Mutual Exec Warns of ‘Hallucination Liability’ in Corporate AI Adoption – AM Best News, accessed on May 24, 2025, https://news.ambest.com/newscontent.aspx?refnum=266304&altsrc=23
11. Cloud Realities Are Slowing AI Ambitions – The New Stack, accessed on May 24, 2025, https://thenewstack.io/cloud-realities-are-slowing-ai-ambitions/
12. Report: Chinese hackers used Cityworks vulnerability to deliver …, accessed on May 24, 2025, https://statescoop.com/chinese-hackers-cityworks-vulnerability-malware-2025/
13. Cityworks Zero-Day Exploited by Chinese Hackers in US Local Government Attacks, accessed on May 24, 2025, https://www.securityweek.com/cityworks-zero-day-exploited-by-chinese-hackers-in-us-local-government-attacks/
14. Chinese hackers breach US local governments using Cityworks zero-day, accessed on May 24, 2025, https://zendata.security/2025/05/23/chinese-hackers-breach-us-local-governments-using-cityworks-zero-day/
15. Cyber News Roundup for May 23, 2025 – RedSeal, accessed on May 24, 2025, https://www.redseal.net/cyber-news-roundup-for-may-23-2025/
16. Computex 2025 Live: Final day wrap-up from Taipei | Tom’s Hardware, accessed on May 24, 2025, https://www.tomshardware.com/news/live/computex-2025
17. CE Pro Weekly Rewind: May 19 – 23, 2025, accessed on May 24, 2025, https://www.cepro.com/news/ce-pro-weekly-rewind-may-19-23-2025/618743/
18. Contracts for May 23, 2025 – Department of Defense, accessed on May 24, 2025, https://www.defense.gov/News/Contracts/Contract/Article/4196970/
19. Hardware Technology & Innovation Breaking News and Press Releases – Business Wire, accessed on May 24, 2025, https://www.businesswire.com/newsroom/industry/technology/hardware
20. Salesforce, Citrix Parent Jockey For Informatica: Report – CRN, accessed on May 24, 2025, https://www.crn.com/news/ai/2025/salesforce-citrix-parent-jockey-for-informatica-report
21. This week at the Capitol: May 19-23, 2025 – CT Mirror, accessed on May 24, 2025, https://ctmirror.org/2025/05/23/ct-2025-legislative-session-may-19-23/