Building on AI's Unfinished Foundation
The opportunities and challenges of AI's platforming process as a general-purpose technology
1The Rapid Rise of Generative AI
Generative AI technology is spreading at an extraordinary pace. Estimates suggest 2.4 billion people worldwide use generative AI platforms monthly. While AI coding platforms like Cursor reached $2 billion revenue run rate by early 2026, Perplexity surpassed 100 million monthly users by challenging the search market. Not just AI-native companies, Salesforce's Agentforce achieved $1.2 billion in annual recurring revenue, showing widespread adoption.
2General-Purpose Technology Vision
AI's true promise isn't just being another successful technology platform but becoming a general-purpose technology like electricity or combustion engines. Such technologies have the potential to transform organizations, industries, and ultimately entire economies. However, current developments remain superficial by these standards. Complementary innovations, organizational integration, and economic transformation processes are still in their infancy.
3The Importance of Platforming
General-purpose technologies need to undergo a platforming process to realize their potential. This means stabilizing technological, industrial, and institutional architectures. As with electricity, the Niagara Falls hydroelectric project (1895-1896) stabilized the technological architecture while public utility commissions established in 1907 completed the institutional framework.
4AI's Current State
AI's platforming process remains incomplete. Technological architecture is forming around pretrained language models, specialized hardware, and cloud infrastructure. Industrial architecture shows division of labor among model providers, application developers, and infrastructure companies. However, institutional architecture remains unstable due to rapidly changing regulations and uncertainties.
5Technological Architecture
AI's technological architecture is coalescing around pretrained foundation models, specialized chip-based hardware infrastructure, and API-mediated delivery. This layered stack (chips, cloud infrastructure, foundation models, applications) forms the ecosystem's foundation. However, significant challenges persist regarding model stability, scalability, and reliability.
6Industrial Architecture
AI's industrial architecture is rapidly evolving. Clear divisions emerged between model providers (OpenAI, Anthropic), infrastructure companies (Nvidia, AWS), and application developers. Yet value capture mechanisms and pricing models remain unsettled. Creating sustainable business models proves particularly challenging for smaller players.
7Institutional Architecture
AI's institutional architecture remains the most unstable area. Regulatory frameworks change rapidly, ethical standards are unclear, and accountability mechanisms lack definition. While initiatives like the EU AI Act and US executive orders exist, no globally harmonized approach has emerged. This uncertainty negatively impacts corporate investment decisions.
8Risks of Unfinished Platforming
The incomplete platforming process creates significant risks. Changing technological standards could jeopardize investments. Uncertainties in industrial architecture threaten business models. Sudden regulatory changes may increase compliance costs. Mid-sized companies are particularly vulnerable to these risks.
9Early-Stage Opportunities
Important opportunities exist even during the platforming process. Specializing in layered architecture (chip design, model training, application development) can be profitable. Demand grows for complementary innovations (data cleaning tools, model management systems). Vertical solutions (AI tailored for healthcare, legal, retail) may offer competitive advantages.
10Corporate Integration Strategies
Companies integrating AI into business processes should adopt modular approaches. API-based solutions maintain flexibility. Balance internal capability building with external sourcing. Starting small with pilot projects then scaling successful ones reduces risks. Cross-functional teams can accelerate adaptation.
11Investment Priorities
AI investments should focus on three areas: 1) Core technology capabilities (data engineering, model understanding), 2) Complementary assets (proprietary datasets, domain expertise), 3) Adaptive organizational structures. Investing in long-term strategic capabilities rather than short-term gains yields more sustainable results.
12Future Perspective
For AI to realize its general-purpose technology potential, the platforming process must complete. Technological standards need to settle, industrial division of labor must clarify, and institutional frameworks require stabilization. Companies should both capitalize on opportunities and manage risks during this transitional period.
Original article: MIT Sloan Management Review — Yapay Zekanın Tamamlanmamış Temelleri Üzerine