Top Tech News Today, February 17, 2026

It’s Tuesday, February 17, 2026, and here are the top tech stories making waves today — from AI and startups to regulation and Big Tech. Artificial intelligence is redrawing the map of global power — not just in software, but in silicon, energy, regulation, and even space.

In the past 24 hours, we’ve seen mounting pressure on chip licensing models, governments racing to secure AI data center dominance, enterprises scrambling to govern runaway AI agents, and fresh cybersecurity warnings that remind us how fragile the digital backbone remains. Apple is tightening its global launch cadence. Cohere is pushing multilingual AI closer to the edge. India is positioning itself as a compute superpower. And quantum research is quietly laying foundations for the next frontier.

What ties these stories together isn’t hype. It’s infrastructure. The AI era is no longer about demos — it’s about power grids, legal frameworks, compute geography, governance controls, and who controls the stack. The winners won’t just build smarter models. They’ll build smarter systems.

Here are the 15 global technology news stories shaping that future today.

Technology News Today

Qualcomm Tech faces a UK licensing lawsuit that could reshape AI-era chip royalties

Qualcomm has been hit with a UK lawsuit that challenges how it licenses key cellular technologies, putting a spotlight on the royalty model underpinning much of the smartphone and connected-device economy. The case lands at a moment when those same radios are becoming foundational to always-connected AI PCs, wearables, and edge devices that need cloud-grade intelligence without relying on Wi-Fi. If courts start tightening what “fair” looks like in licensing, the impact will ripple far beyond phones.

What’s new is the timing: the industry is heading into another platform shift, where AI features are becoming default expectations on consumer hardware, and connectivity is becoming the silent enabler. A licensing reset would influence everything from device pricing to how startups choose modems and reference designs, and it could alter negotiating leverage between chip suppliers, OEMs, and platform giants. It also adds another layer of uncertainty for investors who’ve treated connectivity IP as a predictable cash flow stream, even as AI and on-device compute push margins in new directions.

Why It Matters: A legal challenge to Qualcomm’s licensing playbook could change the economics of connected AI devices across phones, PCs, and emerging hardware categories.

Source: Reuters.

EU opens a formal probe into Shein’s platform design, escalating Tech regulation under the DSA

European regulators have opened a formal investigation into Shein, focusing on illegal products and concerns that its user experience may be designed to be “addictive.” The case is rooted in the EU’s Digital Services Act (DSA), which is increasingly becoming the rulebook that large consumer platforms must follow, not just in Europe but also as a de facto global standard for product safety, transparency, and recommender systems.

For tech and startups, the signal is clear: regulators are expanding beyond content moderation into the mechanics of engagement itself, including reward loops, recommendation transparency, and how platforms handle age-restricted goods. That matters because “growth” has often been optimized through frictionless design, and now those very patterns can trigger enforcement. As the EU presses ahead, the likely outcome is more compliance overhead for any consumer marketplace that touches Europe, plus a stronger incentive for founders to build with auditability and algorithmic transparency from day one.

Why It Matters: The EU is moving from broad rules to high-profile enforcement, and product design choices are now a regulatory risk.

Source: Reuters.

Cohere launches open multilingual AI models, pushing on-device Tech beyond English-first

Cohere has released a new family of open, multilingual models designed to provide broad language coverage while remaining small enough to run on everyday devices. That “small-but-capable” direction matters because it aligns with where the market is heading: companies want AI that is cheaper to deploy, easier to control, and viable in regions where connectivity or cloud costs are constraints. Multilingual capability is also no longer optional; it is product-critical for consumer apps, customer support, education, and local commerce across Asia, Africa, and Latin America.

The competitive significance is subtle but real. Big frontier models keep getting larger, but the distribution is shifting toward models that can run closer to users, with better latency, privacy, and cost. Open releases also accelerate downstream innovation: startups can fine-tune and ship faster without waiting for permission or expensive enterprise contracts. The near-term beneficiaries will be app builders and device makers who want “AI everywhere” without “cloud bills forever.”

Why It Matters: Open, multilingual models lower the barrier for global AI products and make on-device intelligence more practical at scale.

Source: TechCrunch.

Apple Tech schedules a March 4 global event, hinting at a new hardware-and-AI launch rhythm

Apple has teased a March 4 “special Apple experience,” with events centered in New York and parallel gatherings in London and Shanghai. The geographical spread is the story: Apple appears to be leaning into globally synchronized launches, which can tighten media cycles and shorten the gap between announcement and availability across key regions. With Mobile World Congress happening the same week, Apple is effectively inserting itself into the global hardware conversation without waiting for the traditional spring cadence.

While Apple hasn’t confirmed products, the speculation points to multiple device refreshes, including lower-cost and mainstream hardware that could broaden the installed base for on-device AI features. That matters because Apple’s AI positioning is increasingly tied to silicon strategy. If Apple pushes more devices powered by its own chips into lower price bands, it creates more surface area for AI features that are privacy-forward and run locally, and it forces competitors to keep pace on efficiency, not just raw model size.

Why It Matters: Apple’s launch timing and global staging suggest it is optimizing hardware distribution to support its next wave of on-device AI features.

Source: The Verge.

Unity Tech adds AI-driven game creation prompts, accelerating the “vibe coding” era for developers

Unity is expanding its AI-powered creation features that let developers generate game elements from prompts, aiming to reduce the friction between idea and a playable prototype. For indie studios and small teams, this is a meaningful shift: assets, interactions, and early-level layout have historically been time sinks that require specialists. AI tools don’t eliminate craftsmanship, but they compress iteration loops, allowing teams to test more concepts and converge faster on what players actually enjoy.

Strategically, game engines are becoming platforms for AI-assisted production, not just for rendering and physics. That raises new competitive stakes between Unity, Unreal ecosystem partners, and a growing set of AI-native creation startups. It also creates a new learning curve: teams will need governance for licensed assets, provenance, and “what exactly did the model generate?” questions that can become legal and reputational risks. But the direction is obvious: the engine is becoming a co-pilot, and that changes how games get built and how quickly new studios can form.

Why It Matters: AI creation inside the engine compresses development cycles and lowers the cost of experimentation for game startups.

Source: The Verge.

AI data centers shift to higher-voltage power systems, creating a new infrastructure race for suppliers

A major redesign is underway in how AI data centers distribute electricity, with the industry increasingly pushing toward higher-voltage architectures such as 800V systems. The driver is straightforward: AI workloads are power-dense, and legacy approaches waste space and energy while increasing copper usage. Higher-voltage systems can improve efficiency and reduce physical constraints, but they also require new equipment, layouts, and safety and maintenance practices.

This matters for the startup ecosystem because it changes who wins in the supply chain. The beneficiaries are not only hyperscalers. It is also the industrial and semiconductor companies building power electronics, conversion systems, and cooling-adjacent infrastructure. As AI demand continues, the most valuable “AI companies” may not be model labs at all, but the picks-and-shovels players enabling power delivery at scale. Founders building in energy optimization, thermal management, and data center retrofits are moving from niche to essential.

Why It Matters: The AI boom is forcing a power-architecture upgrade that will reshape data center design and create major new markets for infrastructure suppliers.

Source: Financial Times.

India Tech targets up to $200B in AI data center investment, raising the stakes for global infrastructure buildouts

India says it aims to attract up to $200 billion in data center investment as it positions itself as a global AI hub. The pitch is both economic and strategic: AI compute is becoming a national capability, and countries want the jobs, supply chains, and sovereignty that come with hosting the infrastructure. For India, the opportunity is to convert its scale, developer base, and market growth into physical compute capacity that can serve domestic demand and regional customers.

The implications extend beyond India. If capital flows into Indian data centers at the scale being discussed, it reshapes where AI workloads run, where cloud competition intensifies, and how chip and networking supply chains prioritize regional buildouts. It also highlights constraints: power availability, grid stability, water use, and permitting. For startups, India’s data center push can catalyze adjacent markets in energy, cooling, monitoring, and AI governance, particularly as enterprise adoption accelerates.

Why It Matters: If India secures even a fraction of this data center capital, it could shift the gravity of AI infrastructure toward the Global South.

Source: AP News.

Pennsylvania Tech’s $90B AI-and-energy project wave moves forward, spotlighting local pushback and grid pressure

A new tracker details progress across massive AI data center and energy projects in Pennsylvania, showing regulatory milestones, early spending, and the friction points that typically emerge once shovels get close to the ground. The story here is not just the dollar figure. It is a reality that AI infrastructure is becoming a local political issue, with communities questioning land use, energy pricing, and who benefits from the buildout.

For the broader tech ecosystem, this is the pattern to watch: data centers are no longer invisible warehouses. They are a public topic tied to grid capacity, reliability, and costs. That shifts how Big Tech and infrastructure developers approach projects. Expect more commitments around on-site power, local economic packages, and transparency about water and electricity usage. It also creates room for startups that help operators demonstrate efficiency, model community impact, and reduce operational strain through improved power management and cooling systems.

Why It Matters: AI infrastructure is colliding with real-world constraints, and local acceptance is becoming as important as capital.

Source: Technical.ly.

FTC probes Microsoft Tech cloud practices, putting AI platform power under a brighter antitrust lens

The U.S. Federal Trade Commission is investigating Microsoft’s business practices, with attention on cloud and AI-related operations. The scrutiny matters because Microsoft sits at an unusually powerful intersection: enterprise software licensing, cloud infrastructure, identity and security tooling, and now AI distribution through partnerships and platform integration. When regulators look at this stack, the question is not just market share, but how incentives and bundling might shape where AI workloads run and which competitors can realistically challenge the default options.

For startups and enterprise buyers, this kind of probe can have downstream effects even before outcomes are clear. It can push platform companies to loosen terms, improve portability, or adjust licensing practices that make multi-cloud more expensive. It also encourages investors to reassess dependency risk: if your product is built on a single cloud or single AI distribution channel, regulatory shifts can quickly change unit economics and go-to-market assumptions.

Why It Matters: Antitrust pressure on Microsoft’s cloud-and-AI stack could influence pricing, portability, and competitive access across enterprise tech.

Source: The Register.

CIOs warn of “ungoverned AI agents” inside companies, raising the urgency for enterprise controls

A new report highlighted by The Register says a large share of CIOs believe employees are creating AI agents and apps faster than IT can govern them, and many fear sensitive data exposure as a result. This is the inevitable second phase of AI adoption: after experimentation comes the realization that productivity gains can turn into security and compliance liabilities. In practice, employees will continue to build using tools they can access, so governance cannot be purely restrictive. It has to be structured and usable.

For the startup ecosystem, this fuels demand for a new category of enterprise tooling: AI policy enforcement, monitoring, agent permissioning, data loss prevention tailored to model workflows, and audit trails that satisfy legal and compliance teams. It also changes procurement behavior. Companies that once bought “AI features” will increasingly buy “AI controls,” and winners will be the vendors that can integrate with existing security stacks while staying lightweight enough for actual adoption.

Why It Matters: Enterprise AI is shifting from “try it” to “control it,” creating a fast-growing market for governance and security startups.

Source: The Register.

Password manager Tech faces new “malicious server” risk, challenging assumptions about vault safety

Researchers tested leading password managers and found scenarios in which a vault could be compromised if the server a client connects to is malicious. This matters because password managers are core infrastructure for both consumers and businesses, and the security promise is simple: even if someone intercepts traffic, your secrets remain protected. The research suggests there are edge cases where that trust boundary is weaker than users assume, especially as syncing and cross-device convenience have become the default.

The broader implication is that authentication is entering a turbulent period. As passkeys roll out unevenly and AI tools improve phishing techniques and scale, password managers remain a frontline defense. Any perceived weakness can quickly shift enterprise policy and consumer behavior. Expect product changes, including stronger server authentication models, more hardened key handling, and a renewed focus on local-first architectures. It may also trigger uncomfortable questions for vendors: how to communicate nuanced risk without causing a panic that pushes users back to worse habits.

Why It Matters: When password managers are questioned, the security foundation of millions of users and enterprises is at stake.

Source: SecurityWeek.

Eurail confirms stolen traveler data is for sale, raising real-world phishing risk for global travelers

Eurail says data stolen in a recent breach is being sold on the dark web, including personally identifying travel-related information. The immediate danger is not abstract. Travel data is unusually useful for criminals: it can enable targeted phishing, identity fraud, and scams timed to a person’s itinerary. Even without bank data, travelers are often on the move, distracted, and more likely to click a “ticket problem” message that appears urgent and plausible.

For the tech ecosystem, this is another reminder that breaches are increasingly multi-stage: intrusion, exfiltration, confirmation, and monetization. Each stage expands impact. It also raises pressure on travel-tech firms and consumer platforms to adopt stronger segmentation, stricter retention policies, and better incident communications. Startups building identity protection, fraud detection, and breach response tooling will find receptive enterprise buyers, especially in sectors where personal data is inherently “actionable” to attackers.

Why It Matters: Travel data is highly valuable to targeted scammers, and confirmed resale turns a breach into a live risk for real people.

Source: TechRadar.

UK space Startup Orbex shows off its Prime rocket as insolvency threatens Europe’s launch ambitions

Orbex, a British space-launch startup, has released new imagery of its Prime rocket as it faces potential wind-down through insolvency proceedings. The story is a sharp snapshot of the European launch market reality: technical progress does not guarantee financing continuity, and delays can be fatal when capital markets tighten. Europe wants domestic launch capability for sovereignty reasons, but the path is cluttered with schedule risk, expensive hardware, and limited near-term revenue.

For founders and investors, Orbex is a case study in timing and capital structure. Space is still a sector where the “minimum viable product” is expensive, and the market often demands proof at the hardest stage: integrated hardware, not slides. If Orbex unwinds, it could reshape partnerships, talent flows, and acquisition opportunities, while reinforcing a broader lesson for frontier startups: milestones and cash runway must be aligned with brutal realism, not aspiration.

Why It Matters: A launch startup’s collapse would underscore how hard it is to finance space hardware in today’s market, even with strategic demand.

Source: Aviation Week.

Luma AI shifts compute to Saudi Arabia as “monstrous” AI chip supply comes online

A new report says video AI startup Luma AI is moving compute to Saudi Arabia, drawn by growing access to advanced AI chips and the infrastructure to run them. This is part of a wider trend: compute is becoming geopolitically distributed, and countries with capital and energy advantages are trying to turn those inputs into AI gravity. For startups, the decision is pragmatic. If you can secure reliable accelerators and predictable operating costs, you can ship faster and train bigger models without being bottlenecked by the same queues everyone else faces.

For the global tech ecosystem, this also raises harder questions. Where AI workloads run affects data residency, regulatory exposure, and customer trust, especially for enterprises and governments. It also reorders cloud competition. If regional compute hubs become credible alternatives, hyperscalers may face new pricing pressure or partnership demands. Expect more startups to treat compute location as a strategic lever, not an afterthought, particularly for video and multimodal AI, where costs scale quickly.

Why It Matters: AI compute is becoming a location strategy, and chip availability is starting to pull startups toward new global infrastructure hubs.

Source: Semafor.

New experimental work probes half-integer thermal conductance in quantum Hall states

A newly published Nature Communications paper reports theoretical and experimental results involving half-integer thermal conductance plateaus realized using integer quantum Hall states, leveraging a specialized bilayer graphene geometry. While this is not “product-ready” quantum computing, it is foundational physics research that eventually informs future quantum devices and measurement techniques. Progress in quantum information often depends on improving control, validation, and understanding of exotic quantum behaviors under real experimental conditions.

Why it matters now is the compounding effect. As governments and companies pour money into quantum roadmaps, the bottleneck is frequently not ambition but reliability: can you verify quantum states, manage equilibration, and build systems whose behavior is predictable enough to scale? Advances in quantum transport and thermal measurement add to the toolkit. Over time, these results can influence how researchers design confined geometries, interpret thermal signals, and refine architectures that support more stable quantum operations.

Why It Matters: Breakthroughs in quantum measurement and behavior validation are the quiet enablers that make scalable quantum tech possible in the future.

Source: Nature Communications.