How AI Is Changing Enterprise

Strategic Positioning of AI Startups vs. Model Providers

• The "ChatGPT wrapper" is a flawed business model; long-term value lies in building comprehensive software applications with proprietary business logic and data, not just token output.
• Startups must differentiate their value proposition from what AI models will natively incorporate; the risk is not model capability but being "folded in" by model providers' own consumer-scale applications.
• Pure-play model companies are increasingly difficult to sustain as business models; successful entities function as software companies where the model is an abstracted engine for delivering specific outcomes (e.g., security, compliance, SLAs).
• The "wedge theory" remains valid for startups: begin with a simple product solving a specific workflow, then expand, rather than attempting to build a monolithic platform immediately.
• Enterprises prioritize "outcomes" (e.g., automated contract workflows, EHR transcription) over "models," viewing improved model intelligence as a benefit that allows for better outcomes with less custom hacking.

Market Economics and Pricing Dynamics

• Jevons Paradox is the governing economic principle for the AI revolution: automating tasks lowers costs, which increases consumption and ultimately builds more, leading to abundance rather than jobless scarcity.
• As intelligence becomes commoditized, the cost of tokens is expected to converge toward the cost of bare metal (GPU), similar to the historical convergence of storage pricing among hyperscalers.
• Business models are shifting from fixed annual contracts to usage-based models, particularly for functions replacing BPOs or labor, allowing for elastic scaling without the friction of hiring or infrastructure build-out.
• Gross margins for AI software companies are expanding rapidly (e.g., from 30% to 80%) as token costs decrease, mirroring the trajectory of SaaS storage companies where the underlying compute is a negligible fraction of the delivered value.
• The Total Addressable Market (TAM) for software is projected to expand five-fold over the next decade, not by replacing labor budget, but by funding work that enterprises previously could not afford to do (e.g., global translation, contract data extraction).

Enterprise Adoption and Internal Strategy

• Fortune 500 leaders are moving from "cloud skepticism" to "AI-native" strategies, recognizing that failing to adopt AI creates a competitive disadvantage regarding workforce productivity and speed.
• The "Context vs. Core" framework dictates that companies should buy external software for "context" functions (HR, CRM, ERP) and build homegrown solutions only for "core" IP-generating activities (e.g., wealth management algorithms, drug discovery).
• While CTOs and AI leads care deeply about model specifics (e.g., DeepSeek vs. Anthropic), line-of-business executives and end-users remain indifferent to the underlying tech, focusing solely on workflow execution and accuracy.
• Enterprises are rapidly adopting internal AI tools for coding productivity, knowledge management (interrogating HR data), and customer support, though custom-built chatbots for general knowledge are viewed as a temporary phase before a return to dashboard-based interfaces.
• Security concerns regarding hosted models are diminishing as the cloud infrastructure established over the last 15 years has created a precedent for trust, compliance, and data governance.

Workforce and Societal Impact

• The workforce is becoming "AI-native," with new hires accustomed to using AI for information retrieval and problem-solving, forcing enterprises to modernize technology stacks to retain talent.
• Efficiency gains from AI will likely be reinvested into growth (hiring sales, support, and R&D) rather than retained solely as profit, driven by competitive market pressures.
• The "Black Mirror" scenario of mass unemployment is countered by the potential for lower costs to improve accessibility in healthcare and education, lifting lifestyles across underserved communities.
• The next wave of enterprise adoption involves "agentic" workflows where AI chains together multiple agents to complete complex business processes, moving beyond simple chat-based assistance.
• Open-source models are creating a symbiotic commercial ecosystem where enterprises can use free software while paying for enterprise-grade support, governance, and hosting features.