Article 2: AI’s next investment cycle belongs to applications

Why in news: The AI industry is shifting from infrastructure hype to application-led profitability, highlighting investor focus on real revenues, enterprise adoption, and sustainable business models, similar to how the internet was monetised through applications rather than raw capacity.

Key Details

• Profitability, not performance, is now the key AI challenge
• Massive infrastructure spending has not ensured sustainable returns
• Foundation models face high inference costs and intense competition
• AI applications account for the majority of generative AI spending
• Enterprises are deploying AI at scale, not just experimenting
• Revenue concentration is emerging in a small number of successful AI products
• Investors favour real customers over speculative technology
• Departmental AI, especially coding tools, creates measurable value
• Applications drive demand for models and infrastructure
• Vertical-specific AI solutions offer the strongest long-term potential

AI at a Turning Point

• The AI industry is moving from a phase of heavy investment to one focused on profitability
• The debate is no longer about whether AI works, but whether it can generate sustainable returns
• Future success depends on practical AI applications, not bigger models or more GPUs

AI Infrastructure vs AI Applications

• In 2025, around $320 billion was spent globally on AI infrastructure
• Despite massive spending, foundation model companies operate on thin margins
• High inference costs and intense competition keep profits low
• Even with $13 billion in annualised revenue, OpenAI recorded a $5 billion loss in 2024
• This model relies heavily on venture capital and corporate funding, making it unsustainable long-term

Why AI Applications Are Winning

• Spending on AI applications reached $19 billion in 2025
• Applications accounted for over half of all generative AI spending
• This represented more than 6% of the global software market, just three years after ChatGPT’s launch
• Companies have moved from experimentation to large-scale adoption
• Over 10 AI products now earn more than $1 billion in annual recurring revenue
• Around 50 products generate over $100 million annually

Signals from the Investment Market

• Meta’s $2 billion acquisition of Manus highlights the shift toward applications
• Manus reached $125 million in annual revenue within nine months by delivering task-oriented AI, not just conversational tools
• Investors increasingly prioritise real customers and revenue
• By Q3 2025, there were 265 private equity deals in AI applications
• 65% growth in such deals compared to the previous year
• 78% were add-on acquisitions strengthening existing businesses
• Strategic M&A activity in AI rose sharply, with deal values up 242% year-on-year

Where Real Value Is Emerging

• The departmental AI market is becoming the most valuable segment
• In 2025, AI coding tools made up $4 billion of the $7.3 billion departmental AI market
• Around 50% of developers now use AI coding tools daily
• In top-performing firms, usage rises to 65%
• Major acquisitions, such as ServiceNow–Moveworks, focus on business outcomes, not infrastructure

Applications Drive Models, Not the Reverse

• Anthropic now controls 40% of enterprise LLM spending
• Its rise is driven by dominance in coding applications, with a 54% market share
• OpenAI’s enterprise share has declined despite early leadership
• Applications pull demand for infrastructure and models
• Generative AI reached a 34% contribution margin in 2025, its first profitable year
• Margins could rise to 67% by 2028 as efficiency improves
• Most profits flow to end-to-end solution providers, not raw compute sellers

What Investors Should Focus On

• The next wave of value will come from specific use cases, not generic AI interfaces
• High-value opportunities lie in healthcare, law, finance and manufacturing
• Successful products are deeply embedded in workflows
• Use of proprietary data and operational integration creates defensible businesses
• Core metrics like revenue, retention, growth and profitability matter again
• Circular financing between big tech firms distorts true demand
• AI applications generate external revenue, breaking this cycle

Policy and Regulatory Challenges Ahead

• Governments must address competition concerns as model providers build their own applications
• Smaller application developers face pressure from vertically integrated giants
• Copyright and training data issues are becoming central legal concerns
• Privacy frameworks must adapt to AI agents handling sensitive data
• Over-regulation could slow innovation at the application layer
• Strong merger reviews are needed to prevent anti-competitive acquisitions
• The rise of acqui-hires risks harming innovation and workforce stability

The Bigger Lesson

• The Internet was monetised through applications, not bandwidth
• AI will follow the same path
• Long-term value lies in useful, scalable applications, not just powerful technology

Conclusion

The AI industry is entering a decisive phase where profitability and real-world impact matter more than scale. While infrastructure enabled rapid progress, applications now drive value, revenues, and adoption. Sustainable growth will come from use-case-focused, workflow-embedded AI solutions, just as the internet ultimately succeeded through applications rather than raw bandwidth.

EXPECTED QUESTIONS FOR PRELIMS:

Consider the following with reference to Microsoft Azure:

1. Microsoft Azure is a cloud computing platform that provides services such as computing power, storage, networking, and databases over the internet.
2. Azure follows only the Infrastructure as a Service (IaaS) model and does not support Platform as a Service (PaaS) or Software as a Service (SaaS).
3. Microsoft Azure supports hybrid cloud solutions, allowing integration of on-premises infrastructure with cloud services.

How many of the above is/are correct?

1. Only one
2. Only two
3. All three
4. None

Answer: b