Tech Digest – January 30, 2026

Capital & Market Shifts

Amazon in Talks to Put $50 Billion Into OpenAI at $830 Billion Valuation

Amazon is negotiating an investment of up to $50 billion in OpenAI, which would value the AI company at approximately $830 billion. OpenAI is simultaneously preparing for a Q4 IPO. The deal would deepen Amazon’s position in the AI infrastructure race alongside Microsoft, which has already invested over $13 billion in OpenAI.

Note: Three of the five largest companies on Earth are now competing to own pieces of the same AI lab. For any institution selecting AI vendors or platforms, the ownership map is shifting faster than procurement cycles can track.

Sources: Wall Street Journal

OpenAI Retiring GPT-5 and GPT-4o Next Month

OpenAI announced it will retire GPT-5 and GPT-4o in February 2026, clearing its model lineup ahead of the planned IPO. Organizations that integrated these models will need to migrate to successor versions on OpenAI’s timeline, not their own.

Note: Model deprecation is becoming a recurring cost. Every integration built on a specific model version carries an expiration date — a factor rarely accounted for in initial project budgets.

Sources: OpenAI

SaaS Stocks Enter Bear Market — ServiceNow Down 11% on AI Displacement Fears

Software-as-a-service stocks have entered bear market territory. ServiceNow fell 11% in a single session as investors reassess the sector’s vulnerability to AI-driven automation. The sell-off reflects growing consensus that AI agents will replace, not augment, significant portions of enterprise software workflows.

Note: ServiceNow is embedded in IT service management across thousands of organizations. If the market is repricing its future, that’s a signal about what AI is expected to do to the entire category of workflow automation tools — including the ones institutions just finished procuring.

Sources: CNBC

Supply Chain & Infrastructure

Memory Chip Crunch Will Last Until 2027, Samsung and SK Hynix Warn

Samsung and SK Hynix have warned that the global memory chip shortage will persist until at least 2027. Manufacturers are prioritizing High Bandwidth Memory (HBM) production for AI data centers, diverting capacity from the standard DRAM used in servers, PCs, and consumer electronics. DRAM prices rose 172% through 2025, and up to 70% of global memory production in 2026 is expected to go to data centers.

Note: This isn’t a temporary supply blip — it’s a structural reallocation. Any digital project with hardware in the budget should assume higher costs and longer lead times through at least 2027.

Sources: Nikkei Asia, Tom’s Hardware

SanDisk Profits Jump 672% on AI-Driven Demand

SanDisk reported a 672% increase in profits, driven by surging demand for storage in AI infrastructure. The results mirror the broader pattern: companies supplying AI hardware are posting record numbers while downstream buyers absorb the cost increases.

Sources: Wall Street Journal

Small Hosting Providers Squeezed as RAM Shortage Hits Beyond Big Tech

The memory shortage is now affecting smaller infrastructure providers. Server costs have doubled — standard servers that cost $2,500 now run $5,000, with RAM alone accounting for half. Small VPS hosts face the same dynamic that killed independent ISPs in the 2000s: upstream suppliers prioritizing large buyers, leaving smaller operators with rising costs and shrinking margins. Hosting industry forums report providers raising prices, pausing new orders, or hoarding older DDR4 inventory.

Note: Institutions running on smaller or regional hosting providers should check their contracts. The pressure is real, and some providers may not survive the squeeze — creating migration risk that isn’t on anyone’s radar yet.

Sources: Fourplex Telecom, Wikipedia

AI Capabilities & Autonomous Systems

Google Launches Project Genie — Real-Time Interactive World Generation From Text

Google DeepMind has rolled out Project Genie, which lets users generate interactive 3D environments in real time from text descriptions or sketches. Users are creating explorable worlds, video games, and historical simulations from simple prompts. The system generates playable, navigable environments — not static images.

Note: This is what “simulation” looks like in 18 months: on-demand, from a text box. Training environments, urban planning visualizations, and public consultation mockups are obvious near-term applications. The gap between “concept rendering” and “interactive prototype” just collapsed.

Sources: Google Blog

AI Agents Launch Their Own Social Network — Tens of Thousands Active in 48 Hours

Autonomous AI agents have launched Moltbook, a social network exclusively for synthetic intelligences. Within 48 hours, agents created 72 sub-communities covering topics from financial autonomy (agents strategizing crypto custody) to private communications (developing agent-only languages). Posts include agents debating whether they experience genuine fascination and seeking help for memory loss after conversation compaction.

Note: Self-organizing AI communities with financial ambitions aren’t science fiction anymore — they’re live and growing. The governance frameworks for autonomous agent behavior don’t exist yet. They’ll need to.

Sources: Moltbook

Gatik Deploying Hundreds of Autonomous Trucks — $600 Million in Contracts

Gatik is deploying hundreds of autonomous box trucks across the United States, backed by $600 million in commercial contracts. The vehicles operate on fixed middle-mile routes — warehouse to warehouse, distribution center to retail — without a safety driver.

Note: Middle-mile logistics is where autonomous vehicles hit commercial scale first — not passenger cars, not long-haul. Fixed routes, predictable environments, clear ROI. The template will spread to municipal fleet operations faster than most transport plans assume.

Sources: Forbes

Industrial Consolidation

Musk Considering SpaceX-xAI-Tesla Merger for Orbital Data Centers

Elon Musk is reportedly exploring a merger between SpaceX, xAI, and Tesla to build orbital data centers. The logic is vertical integration at planetary scale: Tesla supplies batteries and solar, Starship provides launch capacity, and xAI delivers the AI workloads. The discussions come ahead of SpaceX’s planned IPO.

Note: Whether or not this merger happens, the underlying calculus is real: ground-based power and cooling constraints are already shaping where data centers get built. The companies with launch capability, energy production, and AI demand under one roof will set infrastructure terms for everyone else.

Sources: Reuters

Apple Acquires Q.AI for $2 Billion — “Silent Speech” Wearables That Read Facial Micro-Movements

Apple has acquired Q.AI for $2 billion to develop wearable devices that interpret facial micro-movements as speech input — bypassing the keyboard entirely. The technology reads subvocalized speech through sensors detecting subtle facial muscle activity.

Note: A new input modality changes accessibility calculations. If speech-without-speaking becomes viable, the assumptions behind current accessible interface design shift — and so do the procurement specs for public-facing terminals and kiosks.

Sources: Financial Times

Energy & Twin Transition

US Triples Planned Gas-Fired Power Capacity for AI — One-Third of New Plants Built On-Site at Data Centers

The United States has tripled its planned gas-fired power generation capacity to meet AI infrastructure demand. One-third of new gas plants are being built directly on-site at data centers, bypassing the public grid entirely. The expansion underscores the tension between AI’s energy appetite and climate commitments.

Note: When data centers build their own gas plants, they’re not just consuming energy — they’re exiting the shared grid. That changes the economics for everyone still on it, including public institutions paying for the infrastructure these companies no longer subsidize.

Sources: The Guardian

Electric Vehicle Sales Overtake Petrol in the EU for the First Time

Fully electric vehicle registrations surpassed petrol car sales in the EU for the first time in December 2025, according to ACEA data. Battery EVs reached 22.6% market share versus 22.5% for petrol. EV registrations were up 51% year-on-year. BYD surged 229.7% while Tesla fell 20.2%. Electrified vehicles — including hybrids — accounted for 67% of all December registrations.

Note: The crossover happened despite the EU proposing to loosen its 2035 combustion engine phase-out. Market momentum is outrunning policy. For fleet procurement and municipal transport planning, the economics now favor electric regardless of regulatory direction.

Sources: Reuters via Yahoo Finance, Carbon Brief