AI News – December 29 – January 4: AI Goes to Space and Banks Tremble
2026 starts with a bang: investors wanting to send servers into space to cool them, European banks forecasting 200,000 redundancies, and China responding to san
Welcome to the first edition of AI News in 2026. If you thought the new year would bring calm and reflection, think again. This week marks a brutal acceleration on all fronts: from data centers in orbit to mass layoffs forecast in the banking sector, to new agentic models that write code better than humans.
It's no longer just "tech hype." It's geopolitics, it's macroeconomics, it's the rewriting of the social contract. Here are the 5 stories defining the start of 2026.
1. Data Centers in Space: The New Gold (and Cold) Rush
The New York Times opens the year with a story that seems like science fiction but is pure engineering necessity: investors and Big Tech are seriously exploring the construction of data centers in low Earth orbit.
🔍 What happened: The energy consumption and cooling demands of AI models on Earth have become unsustainable. Space offers two infinite resources: 24/7 solar energy and extreme cold to dissipate server heat. Pilot projects are already in advanced study phases to move model training (the most energy-intensive phase) outside the atmosphere.
💡 Why it matters: We have reached the physical limit of terrestrial infrastructure. Power grids can't handle it, water for cooling is scarce. Moving infrastructure to space is not just a technical solution, but a geopolitical one: whoever controls the data orbit, controls AI. It's the new space race, but instead of planting flags, we're planting GPUs.
🎯 Our take: It seems crazy until you look at the emissions numbers (80 million tons of CO2 forecast for 2025). AI is forcing humanity to become a multi-planetary civilization not for exploration, but for computational survival.
Source: The New York Times
Also read: AI and Climate: Can Artificial Intelligence Save the Planet?
2. Earthquake in European Banks: 200,000 Jobs at Risk
Computerworld sounds the alarm: European banks could cut up to 200,000 jobs by 2030 due to the massive adoption of AI.
🔍 What happened: It's no longer just chatbots for customer service. AI is replacing entire back-office departments: compliance, risk management, and even internal coding. Banks, squeezed between reduced margins and fintech competition, see automation as the only path to profitability.
💡 Why it matters: This is the first major macroeconomic signal of AI's employment impact in the service sector ("white-collar workers"). Until yesterday we talked about "productivity increase"; today we talk about structural replacement. The banking sector is just the canary in the coal mine.
🎯 Our take: The narrative "AI won't replace you, but a person using AI will" is cracking. In certain sectors, AI will replace you, period, if your job is processing information according to fixed rules. Reskilling becomes a social emergency, not a corporate benefit.
Source: Computerworld
Also read: Predictive Economics: if AI could anticipate a financial crisis
3. OpenAI and Google: The War of Agents and Billions
While we were toasting, the labs didn't stop. NeuralBuddies and Distill Intelligence report a flurry of updates redefining the state of the art.
🔍 What happened:
- OpenAI: SoftBank invests 40 billion dollars. Launched GPT-5.2-Codex, an agentic model specialized in coding and cybersecurity that beats every record on benchmarks. There's also talk of ads in ChatGPT.
- Google: Responds with Gemini 3 Flash and new APIs to integrate "deep research agents."
- xAI: Elon Musk announces a partnership with the US military and a massive expansion of computing capacity.
💡 Why it matters: We have entered the era of Autonomous Agents. We no longer ask the chatbot to write us a function; we ask it to "secure the server" and it does it. SoftBank's monster investment confirms that the financial bubble hasn't burst, it has just concentrated on the winners.
🎯 Our take: The market fragmentation predicted by CNBC is happening. On one side, the generalist giants (OpenAI, Google) become almost state-like infrastructures; on the other, hyper-specialized vertical models are born (like DeepSeek with the new mHC architecture).
Sources: NeuralBuddies, Distill Intelligence, Google Blog
Also read: Invisible Competitors: How to Identify AI Threats Before They Arrive
4. School and AI: The Great Confusion
The New York Times and eSchoolNews paint a chaotic picture of education in 2026. The use of AI has exploded (+15% in a year), but rules are lacking.
🔍 What happened: There's a race by Big Tech to bring chatbots into schools in "digital-first" countries like Estonia and Iceland. But teachers are divided: is AI a personalized tutor that democratizes learning or a crutch that atrophies critical thinking? The RAND survey shows that use is growing much faster than guidelines.
💡 Why it matters: We are raising the first "AI-native" generation. If we delegate the learning and writing process to AI without teaching critical thinking, we risk creating a future workforce incapable of reasoning autonomously. School has become the most delicate front of the technological revolution.
🎯 Our take: We can't ban AI in school (it would be like banning calculators), but we must change what we teach and how we evaluate. The take-home essay is dead; long live oral debate and logic in the classroom.
Sources: NYT, eSchoolNews
Also read: AI and Education: how to teach in the age of technology
5. Chip Geopolitics: China Responds
Reuters reports the blockbuster debut (+80%) of Chinese chipmaker Biren on the Hong Kong stock exchange.
🔍 What happened: While the US tightens sanctions on Nvidia chip exports, China accelerates technological independence. Biren is Beijing's answer to American GPUs. The success of the IPO shows that Asian markets are betting heavily on technological autarky.
💡 Why it matters: AI is not just software, it's hardware. And hardware is geopolitics. The fragmentation of the world into two technological blocs (USA vs China) is solidifying. This will have enormous impacts on global supply chains and future AI standards.
🎯 Our take: Prepare for a 2026 of "Computational Cold War." Companies will have to choose which side to be on, not just politically, but technologically (use American or Chinese APIs?).
Source: Reuters
Also read: AI and Geopolitics: the new digital arms race
📊 What these developments really tell us: The Year of "Brutal Sobriety"
If 2023 was the year of magic and 2024 that of exuberance, 2025 ended with a cold shower of reality. The signals coming from this first week of 2026 — from space data centers to banking layoffs — allow us to draw a precise map for the year that has just begun.
We are not facing a technological slowdown, but an economic and structural metamorphosis. Here are the three vectors that will define the next 12 months.
1. The End of "AI Tourism" and the Beginning of Engineering
CNBC forecasts and CEO comments in Fortune converge on one point: the time for endless "pilot projects" is over. In 2026, AI will no longer be judged for its "wow" potential, but for its impact on gross operating margin (EBITDA). The preemptive layoff of 200,000 bankers is not a speculative move; it's the result of cold calculations on efficiency. Companies will stop "playing" with generative AI to create fun images and start integrating it into "boring" but vital processes: invoice reconciliation, legal analysis, supply chain. The prediction: We will witness a wave of failures among "wrapper" startups (that just resell OpenAI's APIs) and the consolidation of those offering vertical solutions that solve specific industrial problems.
2. The Dimensional Paradox: Giants in Space, Models in Your Pocket
There's a fascinating schizophrenia in this week's news. On one hand, infrastructure becomes so massive and energy-hungry that it has to be shipped into orbit (Space Data Centers). On the other, the market is pushing towards Small Language Models (SLMs). Why? Because the economic and ecological sustainability of Large Language Models (LLMs) is at its limit. In 2026, companies will understand that you don't need an omniscient model with a trillion parameters to summarize an email. A small, specialized model running locally on a laptop (Edge AI) is enough. The prediction: The future is hybrid. We will use "space giants" for scientific research and major global problems, but we will use "pocket-sized" and private models for daily work, reducing latency and costs.
3. The "AI Slop" and the Value of Human Truth
Euronews coins the term "AI Slop" to describe the flood of mediocre, synthetic, and unverified content that is suffocating the web. If AI can generate a million articles per minute, the value of generated content plummets to zero. This creates an economic paradox: in a world of artificial abundance, scarcity becomes human-curated truth. The confusion in schools (reported by the NYT) stems precisely from this: we no longer know how to distinguish signal from noise. The prediction: The "Human Brand" will become a premium asset. We will see the birth of certified "digital oases" (newsletters, closed communities) where the absence of AI or its rigorous curation will be sold at a high price. The ability to filter will become more lucrative than the ability to create.
4. Silicon Geopolitics: A Two-Speed World
The success of China's Biren on the stock market is not just financial news; it's confirmation that the "decoupling" between West and East is irreversible. We will not have a single "Global Artificial Intelligence." We will have a Western bloc guided by ethical standards and private infrastructure (USA/EU), and an Asian bloc guided by state sovereignty and proprietary hardware. This will force multinational companies to make difficult choices: which technological stack to adopt? Where to store data?
In summary: 2025 shook us, showing us that AI is not a toy. 2026 will force us to rebuild the foundations: of work (which is changing), of infrastructure (which is going to the sky), and of truth (which must be defended). We are no longer spectators of a revolution, but architects called to manage its costs and benefits.
Happy New Year to all readers of La Bussola dell’IA.