Summary
The AI industry is experiencing unprecedented acceleration. Dario Amodei from Anthropic predicts that AI systems could completely take over the majority of software development within 6–12 months. In parallel, tech companies are investing massively in hardware ecosystems: Apple is developing a wearable AI pin, OpenAI is launching earbuds, and Anthropic is disrupting classic software with Cloud Cowork. Labor market consequences are already visible – particularly for entry-level professionals. At the same time, pressure on established SaaS providers is mounting, while infrastructure investments like OpenAI's Stargate program cement the oligopoly of big players.
People
- Dario Amodei – CEO of Anthropic
- Demis Hassabis – CEO of Google DeepMind
- Sam Altman – CEO of OpenAI
- Michael Witzenleiter – Host, Private AI
- Jean-Luc Winkler – Co-Host
Topics
- Software development through AI agents
- Wearable AI devices (Apple Pin, OpenAI Earbuds)
- AGI timelines and risk management
- Disruption of classic SaaS models
- AI infrastructure oligopoly
- Labor market shifts
Detailed Summary
The Acceleration of AI Development
At the World Economic Forum 2026 in Davos, leading AI figures presented a radical assessment: Dario Amodei from Anthropic gives self-developing AI systems 6–12 months to take over the majority of software development end-to-end. At Anthropic, developers no longer write code themselves – instead, they give AI models control and only review the results.
This self-reinforcing cycle is the core risk: If AI takes over the development of new AI models itself, innovation cycles shrink dramatically. Time becomes a critical risk variable – regulatory, economic, and societal.
Demis Hassabis from Google DeepMind remains more cautious. He sees progress in coding and mathematics, but emphasizes that scientific breakthrough, creativity, and experimentation remain major hurdles. His time horizon for AGI: Uncertainties until the end of the decade.
Labor Market Effects Already Visible
The first job losses are not occurring among established professionals, but among entry-level and junior roles – classic gateways into working life. Learning becomes more individual and self-directed, but loses income-securing structures. For companies, a dilemma emerges: They must invest in talent, not for today's work, but for tomorrow's potential.
Pressure to act grows exponentially. Five- to ten-year plans risk becoming obsolete. Recruiting, knowledge management, and talent acquisition already show shifts.
Hardware Race: Ecosystem Beats Standalone
Apple is bringing an AI pin similar to a thicker AirTag to market – with two cameras, three microphones, a speaker, and wireless charging. Market launch no earlier than 2027, planned production up to 20 million units.
The secret to success lies not in the device itself, but in integration: Apple leverages ecosystem effects and privacy focus (with support from Google models through a new partnership). At the same time, OpenAI announces earbuds with its own 2-nanometer chip – without an operating system, without an ecosystem. The chances: slim.
Earlier attempts like Humane AI Pin and Rabbit R1 failed due to lack of added value and missing ecosystem integration. Meta and Google have massive advantages here.
Siri Becomes a Generative Chatbot
Apple is fundamentally overhauling Siri. Under the project name "Campus," Apple is integrating a generative AI chatbot deeply into iOS, iPadOS, and macOS. This is late – but potentially disruptive, because Apple could create a stir globally. The cooperation with Google (Gemini models for Siri) opens platforms, but also raises monopolization risks and privacy concerns.
OpenAI: Broad Monetization
Ads in ChatGPT: Initial tests in the US show ads at the bottom of chat responses – for Free and Go users. The pressure is real: OpenAI must cover infrastructure costs (billions per year) through multiple revenue streams. Users might switch to ad-free alternatives like Gemini.
ServiceNow Deal: A 3-year contract that integrates GPT-5.2 directly into IT and customer workflows. Signal to the SaaS industry: Established providers must bet on AI or lose relevance.
Age Verification: OpenAI is rolling out AI-based age prediction. Users can confirm their age via selfie. Child protection vs. data protection – a trade-off.
Stargate Program: By 2029, 10 gigawatts of AI computing capacity should be built – over half already planned. OpenAI is funding energy and network expansion costs in local communities, creating jobs and planning AI academies. The catch: Massive water demand, groundwater risks, and infrastructure oligopoly – only a few players can afford this.
Infinite Memory: Sam Altman announced that AI will in the future be able to permanently retain information and use it across contexts. Massive implications for science and persistent knowledge.
Anthropic and the Cloud Cowork Disruption
Anthropic launches Cloud Cowork – a tool that converts documents, screenshots, notes into reports and tables. This caused stock turbulence:
- Intuit: −16 %
- Adobe, Salesforce: −11 % each
The signal is clear: AI takes over not just add-ons, but core use cases of classic SaaS. But is it a Kodak moment? Probably not structurally yet – the market is overreacting in the short term.
The VC Paradox
Sequoia is investing simultaneously in OpenAI, XAI, and Anthropic (25 billion in Anthropic round). This breaks VC taboos, but clearly signals: They're betting on the oligopoly, not on one winner. "Red and black at roulette" – either one comes or the other.
Quick News
AI Confusion with Dates: The best models confuse 2027 (next year) with 2026 (the year after). This shows: AI prioritizes user satisfaction over factual correctness – a risk when crossing years and at knowledge cutoffs.
Ukraine and AI Alliances: Ukraine is opening its databases – millions of hours of drone footage – to train Western AI models with real combat data. Goal: Improve defense AI.
Core Statements
- Paradigm shift in 6–12 months: AI systems can take over end-to-end software development; innovation cycles shrink radically.
- Labor market under pressure: Entry-level professionals lose first opportunities; income-securing structures are missing.
- Ecosystem beats hardware: Apple AI pin could work; OpenAI earbuds without OS/ecosystem have slim chances.
- SaaS disruption is real: Cloud Cowork and AI agents target core use cases; but structural Kodak moment (not yet) reached.
- Infrastructure oligopoly cements itself: Stargate, Sequoia multi-bets, gigawatt spending – only Big Tech can compete.
- Privacy vs. added value: All new devices/services balance data protection and acceptance; added value decides.
- Time horizon shrinks: Five- to ten-year plans are risky; agile risk management becomes mandatory.
Stakeholders & Affected Parties
| Group | Status |
|---|---|
| Entry-level professionals | Strongly affected – first job losses, loss of orientation |
| Software-as-a-Service providers (Adobe, Salesforce, Intuit) | Threatened by AI agent disruption |
| Tech giants (Apple, Google, OpenAI, Anthropic, Meta) | Benefit massively – market power concentrates |
| Infrastructure-dependent regions | Benefits (jobs, investments) and risks (water, energy) |
| Regulators & politicians | Pressure grows – AGI governance, labor market, data protection |
| Users/Consumers | Gain functionality; lose privacy guarantees |
Opportunities & Risks
| Opportunities | Risks |
|---|---|
| Productivity jump: End-to-end automation in software, R&D, administration | Labor market shock: Massive job shifts, especially junior-level |
| New business models: AI services, agents, consulting | Oligopoly lock-in: Big Tech controls infrastructure & models |
| Local development: AI academies, jobs, infrastructure investments | Resource crisis: Water, energy, groundwater under pressure |
| Healthcare/Science boom: Persistent memory, cross-context learning | Loss of control: Self-reinforcing AI cycles faster than governance |
| Privacy-by-design: Apple, Anthropic promote local/on-device AI | Ads everywhere: ChatGPT, Siri, all interfaces become monetized |
| Leveraging data treasures: Ukraine model for international AI alliances | Militarization: War data trains defense AI; escalation? |
Action Relevance
For Business Leadership:
- Don't delay: Innovation projects must start now; five-year plans are outdated.
- Dual-track: Train employees for future potential (not today); simultaneously advance AI scaling.
- Check ecosystem: Who has platform advantages? (Apple, Google, Meta vs. standalone players.)
- Rethink SaaS: Do core use cases still exist or are they disrupted by AI agents?
For Political Decision-Makers:
- Secure labor market: Retraining programs, income protection for entry-level professionals.
- Infrastructure governance: Regulate water and energy consequences of gigawatt expansion.
- Monopoly control: Sequoia multi-bets, Stargate investments concentrate power – antitrust review needed?
- AGI preparation: Write governance frameworks for self-reinforcing AI cycles.
For Investors & VCs:
- Shorten time horizons: Faster exit cycles, agile portfolios.
- Ecosystem games: Standalone devices lose; platform integration wins.
- Infrastructure reality: Gigawatt investments are not short-term profitable – patience needed.
For Employees:
- Urgent upskilling: Those who stand still get "swallowed" by AI – continuous learning is not optional.
- Entry-level professionals: Use early opportunity or be locked out.