This episode of the AI Daily Brief covers the accelerating compute competition and the emergence of headless agents that are fundamentally changing enterprise software. The discussion features analysis of major announcements from OpenAI, Salesforce, Google, and Microsoft regarding their agentic enterprise strategies.

The episode begins with OpenAI's dramatic scaling of compute infrastructure plans, then explores how companies are redesigning their platforms for agent-first interactions rather than human interfaces. Key developments include Salesforce's Headless 360 platform, OpenAI's workspace agents, Microsoft's Hosted Agents, and Google's Gemini Enterprise Agent Platform.

The conversation examines the business model implications of this shift, particularly how the traditional per-seat SaaS pricing model may give way to consumption-based approaches as agents use software platforms far more intensively than human users ever could.

OpenAI Triples Compute Goals Amid Industry Bottlenecks

OpenAI announced plans to deploy 30 gigawatts of compute by 2030, tripling their previous 10 gigawatt target from the Stargate announcement, representing roughly the peak power demand of New York State.

The company tripled their compute supply in the past year from 0.6 gigawatts to 1.9 gigawatts and claims to have identified more than 8 gigawatts already.

Compute bottlenecks are appearing throughout the supply chain, with Semi-Analysis reporting '100 gigawatts under contract, 10 gigawatts of capacity left through 2030' for power generation equipment.

GE Vernova delivered a massive earnings beat with $17.44 per share versus $1.67 consensus, reporting a $163 billion backlog that likely speaks for their entire decade capacity.

Salesforce Pioneers Headless Enterprise with API-First Platform

Salesforce launched Headless 360, exposing their entire platform including AgentForce and Slack as APIs, MCP tools, and CLI commands without requiring browser interfaces.

Mark Benioff positioned this as a fundamental shift: 'Our API is the UI. All AI agents can access data, workflows, and tasks directly in Slack, voice, or anywhere else.'

Co-founder Parker Harris framed the transformation with the question: 'Why should you ever log into Salesforce again?' as agents handle work through conversational interfaces.

Custom agents on Slack have grown 300% since January, indicating rapid adoption of conversational workflow management.

OpenAI Launches Workspace Agents for Enterprise Teams

OpenAI announced workspace agents that function as 'GPTs on steroids' with enhanced capabilities including memory, scheduling, and team collaboration features.

The agents can operate in the cloud 24/7, work across shared spaces like Slack, and remember past actions and preferences for continuity.

Five internal use cases include software request reviewers, product feedback routers, weekly metrics reporters, outreach agents, and third-party risk managers.

Aaron Levy from Box called this 'probably the biggest news yet in software going headless' that will bring knowledge work agents to the masses.

Microsoft and Google Compete for Agentic Infrastructure Control

Microsoft introduced Hosted Agents in Foundry, providing each agent with its own enterprise-grade sandbox, persistent file system, and support for any orchestration framework.

Satya Nadella emphasized the multi-vendor approach: 'Every agent will need its own computer' with built-in identity, governance, and no vendor lock-in.

Google launched Gemini Enterprise Agent Platform at Cloud Next, with CEO Sundar Pichai calling enterprise agents 'the key to AI monetization efforts.'

Thomas Kurian noted a strategic shift in Vertex AI usage from traditional machine learning to 'a sudden explosion in users building their own custom AI agents.'

Business Model Revolution: From Seats to Consumption

The traditional per-seat SaaS pricing model faces disruption as agents can use platforms 100x more than humans, working 24/7 and processing vastly more data.

Aaron Levy argues that 'agents end up using these underlying platforms far more than people ever did, which opens up use cases that the platform couldn't go after before.'

Debate centers on value capture between horizontal providers like OpenAI/Anthropic, existing systems of record like Salesforce, and new vertical agent companies.

Akash Gupta predicts consolidation: 'Enterprise buyers stop paying $30 a seat for AI in five different products when one ChatGPT seat runs the same workflow.'

The Emerging Agent-Native Software Ecosystem

HubSpot founder Dharmesh Shah argues that successful companies must design 'agentic user experiences' rather than just wrapping existing APIs into MCP servers.

Greg Eisenberg sees 'a trillion dollars up for grabs for agent-first startups' as every SaaS company goes headless within 18 months.

Ivan Burazin suggests coordination tools like Atlassian's Jira will become more valuable as 'agents need standardized project management just like humans do.'

OpenAI's partnerships with Accenture, Capgemini, and PWC highlight the continued need for human expertise in enterprise agent deployment and change management.