🌐 AI & Tech Trends – Week of December 1, 2025
This week didn’t bring big fireworks, but something much more important: direction.
Technology is sending clear signals about where 2026 is heading — less hype, more structure; less promise, more reality.
Europe invests in sovereign infrastructure; thousands of AI projects enter the corporate graveyard; security faces the highest pressure in years; analytics leaves behind the era of infinite dashboards; and Apple reorganizes its internal teams to regain momentum.
These aren’t isolated moves: together they show how AI is maturing and claiming its place as critical infrastructure.
Let’s dive into the most relevant signals.
1️⃣ Europe bets big on its own computing power
📌 Key points:
- The joint announcement by Deutsche Telekom and Schwarz Group to create a mega AI data center in Germany marks a decisive turning point
- Initial estimated investment: €1–2 billion
- Expected energy capacity: several hundred MW, scalable
- Priority: intensive AI workloads, training and sovereign models
What’s happening:
The joint announcement by Deutsche Telekom and Schwarz Group to create a mega AI data center in Germany marks a decisive turning point. This isn’t a political symbol: it’s a strategic shift.
Key project data:
- Initial estimated investment: €1–2 billion
- Expected energy capacity: several hundred MW, scalable
- Priority: intensive AI workloads, training and sovereign models
- Target sectors: industry, telcos, banking, automotive and governments
Europe is catching up:
Some data that explains the urgency:
- 80% of commercial AI computing used by European companies runs outside Europe
- 72% of critical corporate data passes through non-European infrastructure
- Data center demand in the region will grow 17% annually until 2028
- By 2030, AI could consume 3% of European electricity (conservative estimate)
Examples that illustrate the problem:
- BMW and Mercedes have reported weeks of delays for training outside the EU due to privacy regulations
- Nordic governments are migrating workloads to local infrastructure due to sovereignty requirements
Key implications:
- Europe doesn’t compete on speed. It competes on control. And control starts with having its own computing.
- European companies will need to evaluate migrating critical workloads to sovereign infrastructure to comply with future regulations
- An opportunity emerges for local AI infrastructure providers that meet sovereignty standards
Source: “Deutsche Telekom and Schwarz Group to build AI data centre” — Reuters, Nov 30 2025
2️⃣ Hype with a bill: the growing graveyard of AI projects
📌 Key points:
- 60% of AI projects started in 2023–2024 are abandoned or “frozen”
- 80% never reach a stable production environment
- Average cost of a failed project ranges between $250,000 and $2 million
- 1 in 3 analyzed repositories contains AI-generated code that’s poorly maintainable or broken
What’s happening:
The other signal this week is less glamorous: the massive abandonment rate in AI projects.
The numbers are harsh:
- 60% of AI projects started in 2023–2024 are abandoned or “frozen”
- 80% never reach a stable production environment
- Average cost of a failed project ranges between $250,000 and $2 million
- 1 in 3 analyzed repositories contains AI-generated code that’s poorly maintainable or broken
What’s failing?
- Models without owners → no one maintains or updates them
- Duplicate repos → “Frankenstein systems”
- Dirty data → models that predict poorly
- Lack of integration → APIs that change and break pipelines
- Understaffed teams → technical debt impossible to pay
Two quick examples:
- A US retail chain lost $1.4M after abandoning a prediction model that had 9 different versions, all incompatible
- A European bank paused an automatic verification project due to accumulation of more than 70 duplicate scripts and security failures in logs
Key implications:
- The industry is cleaning up. Improvised projects don’t survive the reality of 2026
- Companies must establish clear governance from the start: owners, documentation, integration and maintenance
- The need arises for tools and methodologies to audit and clean up abandoned AI projects
Sources:
“AI project failure rates are on the rise: report” — Cybersecurity Dive, Mar 18 2025
“Why 95% Of AI Pilots Fail, And What Business Leaders Should Do Instead” — Forbes, Aug 21 2025
3️⃣ Security 2026: Zero-Trust stops being optional and becomes mandatory
📌 Key points:
- 52% of companies admit having Shadow-AI (AI used without authorization)
- 41% had AI incidents in the last year
- Regulatory audits will grow 30–50% in 2026 in critical sectors
- Zero-Trust stops being “best practice” and becomes “requirement”
What’s happening:
The attack surface is changing and security teams know it: AI doesn’t just increase capabilities, it also increases risks.
What global reports say:
- 52% of companies admit having Shadow-AI (AI used without authorization)
- 41% had AI incidents in the last year
- Regulatory audits will grow 30–50% in 2026 in critical sectors
- Zero-Trust stops being “best practice” and becomes “requirement”
Emerging new risks:
- Involuntary leaks: prompts containing sensitive data
- Insecure models: open or misconfigured endpoints
- Dangerous logs: confidential information exposed by external tools
- API dependency: one change breaks the entire flow
Real cases:
- A European insurer paused its internal chatbot after detecting customer data mixed in test prompts
- An industrial engineering company exposed internal keys when sending technical queries to an external model without prior filtering
Key implications:
- AI allows moving fast, but also making mistakes fast. And in security, mistakes are expensive
- Companies must implement Zero-Trust controls specific to AI: prompt validation, log auditing, and granular access management
- Security teams need specific training on AI risks and adapted monitoring tools
Source: “The State of AI: Global Survey 2025” — McKinsey & Company, Nov 2025
4️⃣ Enterprise analytics enters a new phase: fewer dashboards, more decisions
📌 Key points:
- 70% of dashboards are not consulted regularly
- 55% of executives waste hours jumping between tools
- Conversational models allow querying data with natural language, detecting anomalies, explaining causes and recommending actions
- Conversational analytics reduces analysis time from hours to minutes
What’s happening:
From OpenText World 2025 and other events, the message is clear: companies are abandoning the obsession with dashboards. They’re adopting conversational analytics where the user asks, and the system responds and acts.
Why it’s happening:
- 70% of dashboards are not consulted regularly
- 55% of executives waste hours jumping between tools
- Conversational models allow:
- Querying data with natural language
- Detecting anomalies in real time
- Explaining causes
- Recommending concrete actions
- Automating reports
Two real examples:
- International logistics: An analyst went from taking three hours preparing a weekly report to three minutes, with automatically generated insights
- Public health: A conversational analytics system detected an abnormal increase in waiting times at three hospitals and generated a staff redistribution plan. Previously, that analysis could take days
Key implications:
- The future isn’t more visualization. The future is more understanding
- Companies must rethink their dashboard investments and migrate toward conversational systems that answer direct questions
- Data analysts will need skills to design and maintain conversational systems instead of just creating visualizations
5️⃣ Apple moves internal pieces: a necessary shift to compete in the next era of assistants
📌 Key points:
- Meta already exceeds 10 million daily interactions with its agent
- Google is integrating Gemini into all its products
- OpenAI and xAI are pushing models that act, not just generate
- A 30% increase in AI investment for 2026 is rumored at Apple
What’s happening:
Apple rarely reveals internal moves, but this week it reorganized key areas of its AI division. This isn’t a minor adjustment: it’s a direct response to competitors’ advances.
The context forcing Apple to move:
- Meta already exceeds 10 million daily interactions with its agent
- Google is integrating Gemini into all its products
- OpenAI and xAI are pushing models that act, not just generate
- The future digital experience will be defined by AI, not the interface
What’s at stake:
- Apple depends on controlling the experience for its ecosystem
- Without robust and contextual AI, that control weakens
- iOS 19 could be the first big bet on hybrid models (on-device + cloud)
- A 30% increase in AI investment for 2026 is rumored
Strategic impact:
Internal changes suggest Apple wants to:
- Open more APIs for developers
- Strengthen on-device AI without sacrificing privacy
- Integrate more contextual agents into native apps
- Regain prominence in the ecosystem’s “intelligent layer”
Key implications:
- Apple doesn’t compete to launch first. It competes to define the standard
- Developers will need to prepare for new Apple AI APIs that allow deeper integrations
- The iOS ecosystem could see a significant renewal with more advanced native AI capabilities
Source: “Nvidia and Deutsche Telekom invest €1 billion in data center” — Techzine.eu, Nov 4 2025
🧭 Conclusion: AI stops promising and starts demanding
The week makes it clear that the industry has entered its adult phase:
- Europe builds its own capacity to not depend on others
- Companies eliminate weak projects to keep what really works
- Security hardens because there’s no room for improvisation anymore
- Analytics evolves to give answers, not graphs
- Apple rearms because the future of the digital experience will be defined by AI
Everything points to a 2026 where AI will be less spectacle and more infrastructure.
Less smoke and more architecture.
Less hype and more responsibility.
AI stopped being a promise. Now it’s structure, security and survival.
📚 Recommended readings
-
Data Centers & Digital Infrastructure
“Data center” — Wikipedia (explica tendencias globales, soberanía digital, demanda energética)
👉 https://en.wikipedia.org/wiki/Data_center -
AI Project Failures & Lessons
“AI winter” — Wikipedia (incluye causas comunes de fracaso, ciclos de hype y abandono de proyectos)
👉 https://en.wikipedia.org/wiki/AI_winter -
Zero Trust Security
“Zero trust security model” — Wikipedia (modelo completo, actualizado, fiable y accesible)
👉 https://en.wikipedia.org/wiki/Zero_trust_security_model
I invite you to comment: Which of these trends do you think will have the greatest impact on your sector? How is your organization preparing for this phase of AI maturity?
See you next week with more updates.
✍️ Claudio from ViaMind
Dare to imagine, create, and transform.
