Investment banking leader Goldman Sachs has released a comprehensive equity research report identifying the next generation of artificial intelligence winners as corporate monetization moves decisively beyond traditional, seat-based software subscriptions. According to the investment bank’s global technology research team, the multi-trillion-dollar artificial intelligence transition is entering a mature operational phase where corporate software revenue is no longer tied to the number of human employees sitting at office desks. Instead, software monetization is shifting toward consumption-based pricing, automated digital agents, and measurable business outcomes.
The research framework, developed by senior technology analysts at the investment bank, outlines how software-as-a-service business models are breaking through historical revenue ceilings. For nearly two decades, enterprise software providers charged flat monthly subscription fees between $20 and $100 per human user. However, as autonomous artificial intelligence agents take over complex customer service, software coding, legal review, and supply chain management, corporations are paying for digital work completed rather than human software access.
The report categorizes the artificial intelligence investment landscape across four distinct phases, identifying standout corporate leaders positioned to capture outsized market share. While early investment cycles rewarded semiconductor chipmakers and basic cloud hosting platforms, the next wave of capital creation will flow directly to enterprise platforms that combine proprietary enterprise data with autonomous workflow execution. Led by enterprise innovators like Salesforce, ServiceNow, Palantir Technologies, Microsoft, and specialized data platforms like Snowflake and Datadog, this business model evolution will expand the global software addressable market from roughly $300 billion toward more than $1.0 trillion over the coming decade.
The Structural Shift from Per-Seat Licensing to Outcome-Based Monetization
The software-as-a-service industry is undergoing its most significant business model transition since enterprise computing migrated from on-premises desktop servers to the cloud fifteen years ago. Under the traditional per-seat licensing model, a software company’s revenue growth was structurally tied to client headcount growth. If a corporate customer hired 10% more employees, the software vendor sold 10% more licenses; if corporate clients executed workforce layoffs, software subscription revenues contracted.
Generative artificial intelligence and autonomous software agents have fundamentally decoupled software revenue from human employee headcounts. Modern artificial intelligence platforms do not act as passive digital tools that wait for a human employee to type commands; they operate as autonomous digital workers that execute complete, multi-step business operations independently.
This capability has forced software companies to abandon seat-based pricing in favor of outcome-based and consumption-metered pricing models.
Under an outcome-based model, software providers charge fees based on verifiable business achievements, such as resolving a customer support ticket without human intervention, auditing a multi-page commercial contract, or generating verified software code.
Goldman Sachs analysts emphasize that outcome-based monetization transforms software from an administrative overhead expense into a direct labor replacement, allowing technology vendors to capture a share of the multi-trillion-dollar global labor market.
Breaking Free from the Limitations of Human Employee Headcounts
The primary economic limitation of the traditional software-as-a-service model was its capped total addressable market. In an economy with a finite number of white-collar office workers, software vendors eventually saturated their potential customer base, forcing companies to rely on price increases to maintain double-digit revenue growth.
Autonomous software agents break through this human ceiling entirely:
- A digital software agent operates continuously 24 hours a day, 365 days a year, processing millions of complex data interactions without taking breaks or requiring human supervision.
- Corporations can deploy 10,000 digital customer service agents during sudden demand spikes and scale back to 1,000 agents during quiet periods, paying strictly for computational work delivered.
- Software vendors can monetize automated workloads that human employees previously had no time to execute, such as real-time fraud monitoring on every financial transaction.
- Enterprise clients willingly pay premium fees for software that directly reduces corporate payroll overhead and eliminates administrative backlogs.
By expanding software value from assisting human workers to replacing manual labor tasks, software platforms are unlocking massive new revenue streams across global enterprise accounts.
Unpacking Goldman Sachs’ Four Phases of the AI Investment Cycle
To help institutional portfolio managers navigate the evolving technology landscape, Goldman Sachs structured the artificial intelligence investment super-cycle into four sequential phases. Each phase represents a distinct layer of the technology stack as value migrates from physical silicon to commercial applications.
The four phases outline a clear investment roadmap:
- Phase One (Semiconductor Infrastructure): Dominated by pure-play chip designers, foundries, and packaging specialists like Nvidia, TSMC, and Broadcom that manufacture the physical hardware required to train foundation models.
- Phase Two (Cloud Infrastructure and Platform Enablers): Led by hyperscale cloud providers and digital infrastructure operators, including Microsoft Azure, Amazon Web Services, Google Cloud, Oracle, and Equinix, that build and power high-density computing campuses.
- Phase Three (AI-Enabled Software and Workflow Applications): Comprising enterprise application platforms like Salesforce, ServiceNow, Palantir, Intuit, and SAP that integrate autonomous agents directly into corporate systems of record.
- Phase Four (Broader Economic Productivity Beneficiaries): Encompassing traditional corporate enterprises across healthcare, banking, manufacturing, and logistics that leverage artificial intelligence automation to expand operating profit margins.
Goldman Sachs notes that while Phase One and Phase Two delivered massive initial returns, market attention is shifting decisively toward Phase Three, where software developers are demonstrating real-world monetization through autonomous agent deployments.
The Rise of Autonomous Software Agents and Consumption Pricing
The core operational breakthrough driving Phase Three monetization is the emergence of autonomous software agents. Unlike early generative chatbots that required continuous human prompt engineering, autonomous agents are goal-driven software programs that perceive digital environments, make independent decisions, and execute complex workflows across multiple software applications.
An autonomous sales agent can research potential corporate leads, draft personalized outreach emails, schedule product demonstrations, answer technical inquiries, and update customer relationship databases without human intervention.
Similarly, an autonomous IT support agent can diagnose network outages, reset employee security credentials, and deploy software patches automatically within seconds.
Software leaders that control proprietary enterprise data are commercializing these agentic capabilities through usage-based and outcome-based pricing frameworks, generating rapid revenue growth.
Salesforce Agentforce and the $2 per Conversation Benchmark
Enterprise customer relationship management leader Salesforce has established the industry benchmark for commercial agentic monetization with the launch of its Agentforce platform. Rather than charging a flat monthly subscription fee per sales representative, Salesforce introduced an outcome-based pricing tier of $2.00 per conversation resolved by an autonomous digital agent.
The economic model behind Agentforce illustrates the power of outcome-based pricing:
- A commercial retail bank or e-commerce platform processing 500,000 automated customer service inquiries monthly generates $1.0 million in monthly software fees for Salesforce.
- The corporate client saves millions of dollars in customer support staffing expenses while providing customers with instant, 24/7 service resolutions.
- Salesforce customers can build, customize, and deploy autonomous agents across sales, customer service, marketing, and e-commerce within hours using low-code visual builders.
- The platform integrates directly with Salesforce’s Data Cloud, grounding autonomous agents in verified customer purchase histories, support records, and corporate knowledge bases.
Goldman Sachs analysts identified Salesforce as a primary Phase Three beneficiary, noting that usage-based agent monetization will accelerate organic top-line revenue growth toward 12% to 15% annually.
ServiceNow’s Pro Plus Automation and Workflow Monetization
Enterprise workflow automation giant ServiceNow has executed one of the fastest artificial intelligence monetizations in the software industry through its Now Assist and Pro Plus subscription tiers. ServiceNow operates as the digital operating system for corporate IT service management, employee onboarding, and customer workflows across more than 85% of the Fortune 500.
ServiceNow’s commercialization strategy combines tiered subscription upgrades with consumption credits:
- The company’s premium GenAI-enabled Pro Plus tier commands a 60% pricing premium over standard enterprise software licenses.
- Now, Assist automated agents resolve up to 50% of routine IT helpdesk and human resources inquiries without human technician intervention.
- The company’s annual contract value for artificial intelligence-enabled software products surpassed $150 million within twelve months of commercial launch, setting corporate growth records.
- ServiceNow integrated specialized autonomous code-generation tools that allow corporate developers to generate customized enterprise workflows using natural language prompts.
By embedding autonomous intelligence directly into existing corporate IT service workflows, ServiceNow enables enterprise customers to capture immediate operational efficiencies, driving high net expansion rates across its global client base.
Palantir’s Artificial Intelligence Platform and Enterprise Expansion
Data integration and analytics pioneer Palantir Technologies has emerged as a dominant leader in enterprise artificial intelligence deployment through its Artificial Intelligence Platform (AIP). Palantir’s software allows commercial corporations and defense agencies to connect large language models directly to operational business data and physical machinery.
Palantir’s commercial acceleration is driven by its innovative go-to-market bootcamps:
- Palantir conducts intensive, multi-day customer bootcamps that allow enterprise engineers to build functional, autonomous workflows on private corporate data within five days.
- United States commercial customer counts expanded by more than 80% year-over-year, driven by rapid enterprise adoption across healthcare, manufacturing, defense, and energy sectors.
- Total commercial contract value surged as corporate clients expanded pilot deployments into multi-million-dollar enterprise-wide software licenses.
- The platform enforces strict ontological security controls, ensuring that autonomous software agents operate within verified corporate policy boundaries and audit logs.
Goldman Sachs research highlights Palantir’s unique competitive position, emphasizing that its deep integration with mission-critical defense and industrial operations creates an insurmountable barrier to entry for competing software startups.
Foundational Cloud Platforms and Data Infrastructure Gatekeepers
While enterprise application vendors lead Phase Three monetization, foundational public cloud providers and specialized data infrastructure platforms continue to capture immense value by providing the underlying computing, storage, and data preparation layers required to run autonomous agents.
An autonomous artificial intelligence agent cannot function without access to clean, structured, and continuously updated corporate data.
Before an enterprise can deploy a customer-facing agent, it must consolidate fragmented data from dozens of disconnected databases, clean unstructured document archives, and establish real-time data pipelines.
This data preparation requirement has generated a massive commercial tailwind for cloud data platforms and observability providers that manage the foundational plumbing of the digital economy.
Microsoft Azure, Amazon AWS, and Google Cloud Scaling Consumption
The Big Three hyperscale cloud providers—Microsoft Azure, Amazon Web Services, and Google Cloud—are capturing the primary revenue stream generated by autonomous model execution. Because autonomous agents generate multiple internal reasoning loops and query external databases repeatedly to solve complex problems, they consume significantly more computing bandwidth than standard conversational search queries.
The hyperscalers are monetizing this expanding consumption wave:
- Microsoft Azure’s consumption-based infrastructure revenue expanded by 33% annually, surpassing a $100 billion annual sales run rate driven heavily by OpenAI model hosting and enterprise Copilot queries.
- Amazon Web Services scaled its Bedrock platform, providing enterprise clients with managed access to top-tier foundation models while integrating custom Trainium2 silicon to lower compute costs by 40%.
- Google Cloud Platform expanded annual revenues past $45 billion, powered by deep enterprise integration of its multimodal Gemini foundation models and specialized Tensor Processing Units.
- Cloud providers are monetizing specialized software development tools, vector databases, and API management gateways that corporate developers require to build custom agent swarms.
The transition from human software seats to autonomous agents ensures that cloud infrastructure consumption will continue to expand exponentially, providing durable multi-year growth for hyperscale platforms.
Snowflake, Datadog, and the Multi-Billion-Dollar Data Preparation Layer
Specialized data warehousing and cloud observability platforms represent the critical intermediaries enabling enterprise artificial intelligence adoption. Companies cannot deploy effective artificial intelligence agents if their underlying corporate data is inaccurate, unindexed, or vulnerable to security breaches.
Leading infrastructure gatekeepers are capturing expanding enterprise budgets:
- Snowflake: Expanding its Data Cloud platform to provide native hosting for large language models through Cortex AI, allowing corporate clients to run generative models directly inside secure data warehouses without exporting sensitive files.
- Datadog: The cloud observability leader expanded its platform to monitor artificial intelligence application performance, tracking token consumption, model latency, prompt drift, and software errors across distributed computing clusters.
- Dynatrace and AppDynamics: Providing automated root-cause analysis that allows IT administrators to monitor autonomous agent workflows and prevent system crashes in real time.
- MongoDB and Confluent: Scaling real-time data streaming and vector search capabilities, enabling autonomous agents to access live operational data feeds instantly.
Goldman Sachs emphasizes that data infrastructure providers command high gross profit margins exceeding 75% to 80%, offering investors defensive, high-quality exposure to the artificial intelligence boom without taking direct foundation model obsolescence risk.
Physical Infrastructure Enablers: Power, Cooling, and Custom Silicon
The expansion of artificial intelligence monetization beyond software seats is supported by a massive industrial infrastructure buildout. High-density graphics processor clusters and continuous autonomous agent workloads consume staggering amounts of electrical power and generate intense thermal heat that legacy data center facilities cannot support.
Furthermore, technology giants are designing proprietary custom application-specific integrated circuits (ASICs) to lower computing costs and reduce their reliance on expensive merchant graphics processors.
Specialized semiconductor design houses and industrial thermal management conglomerates are capturing multi-billion-dollar backlogs by providing the physical and silicon components required to keep the global computing engine running.
Broadcom and Marvell Capturing the Custom ASIC Semiconductor Wave
While Nvidia maintains a dominant market share in general-purpose graphics processors for foundation model training, hyperscale cloud providers are investing billions of dollars to develop custom silicon tailored to their internal workloads. Custom application-specific integrated circuits offer superior power efficiency and lower per-token operational costs for continuous inference workloads.
Semiconductor design leaders Broadcom and Marvell Technology dominate this custom silicon market:
- Broadcom co-designs custom artificial intelligence accelerators for leading cloud titans, including Google’s Tensor Processing Units and Meta’s MTIA processors, generating tens of billions of dollars in specialized semiconductor revenue.
- Marvell Technology expanded its custom silicon design pipeline, partnering with major cloud providers to develop high-speed optical interconnects and custom compute engines.
- Custom silicon processors reduce total cost of ownership by 30% to 50% for high-volume, standardized inference workloads, enabling cloud operators to offer affordable pricing for autonomous agent deployments.
- Broadcom’s high-speed Ethernet switching silicon, led by the Tomahawk and Jericho platforms, commands a dominant market share in connecting tens of thousands of custom accelerators into unified computing meshes.
Goldman Sachs identified Broadcom as a premier Phase One and Phase Two beneficiary, highlighting its unmatched technical moat in custom semiconductor design and high-speed networking.
Vertiv and Eaton Modernizing Multi-Kilowatt Data Center Grids
The physical heat generated by high-density computing racks has initiated a complete revolution in electrical distribution and thermal cooling engineering. Modern server cabinets housing advanced processors consume between 50 kilowatts and 120 kilowatts of continuous power per rack, generating thermal loads that traditional air-cooling fans cannot dissipate.
Industrial engineering leaders Vertiv Holdings and Eaton Corporation are capturing record order backlogs:
- Vertiv Holdings commands a dominant global market share in high-density direct-to-chip liquid cooling systems, manufacturing specialized coolant distribution units, micro-channel cold plates, and facility chillers.
- Eaton Corporation supplies prefabricated modular electrical substations, medium-voltage switchgear, and uninterruptible power supply systems designed specifically to handle volatile, high-density computing loads.
- The global data center liquid cooling market is projected to expand at a compound annual growth rate of 16.1%, climbing from $13.23 billion toward $37.62 billion by 2033.
- Industrial manufacturers possess strong pricing power and long-term order backlogs extending two to four years into the future, providing exceptional earnings visibility.
Investing in physical power and cooling champions provides institutional investors with tangible, asset-backed exposure to the artificial intelligence buildout, insulated from software valuation volatility.
Strategic Implications for Enterprise Software and Global Investors
The evolution of artificial intelligence monetization beyond software seats carries profound strategic implications for institutional asset allocators, venture capital funds, and corporate executives. The historical metrics used by Wall Street to value software companies—such as annual recurring revenue growth per employee seat, customer retention rates, and customer acquisition costs—are undergoing a fundamental recalibration.
Investors must now evaluate software companies based on their access to proprietary data assets, the efficiency of their autonomous agent orchestration, and their ability to execute consumption-based and outcome-based billing.
Companies that successfully adapt to this new paradigm will experience compounding multi-year revenue growth, while legacy software vendors that remain trapped in static, seat-based licensing will face structural decline.
Expanding Total Addressable Market from $300 Billion Toward $1 Trillion
The transition to outcome-based monetization fundamentally expands the total revenue potential of the global software industry. In traditional economics, software spending represented a modest 3.0% to 5.0% slice of total corporate IT budgets, while human labor compensation accounted for more than 60% of total enterprise operational expenditures.
By providing autonomous digital agents capable of executing labor-intensive business processes, software platforms are capturing capital previously allocated to corporate payrolls:
- The global enterprise software addressable market is projected to expand from roughly $300 billion today toward more than $1.0 trillion by 2035.
- Software providers with established systems of record will capture expanding gross profit margins as automated agents handle millions of routine corporate tasks at negligible marginal cost.
- Mid-market commercial enterprises can scale business operations and enter international markets without executing massive, expensive hiring campaigns.
- Enterprise clients will consolidate software spending with platform gatekeepers that offer integrated agentic ecosystems, squeezing fragmented single-feature point solutions out of the market.
This massive market expansion provides long-term institutional investors with compelling opportunities to compound capital across leading Phase Three software compounders.
The Long-Term Horizon for Autonomous Enterprise Operations
Looking toward the end of the decade, the corporate enterprise will evolve into an autonomous, self-optimizing digital organism. The future of business operations is not human workers manually logging into separate software applications to update records; it is swarms of specialized artificial intelligence agents collaborating autonomously across private enterprise networks.
Key structural trends that will define the next decade of enterprise software include:
- Multi-Agent Orchestration: Autonomous agents from different software platforms collaborating seamlessly—such as a Salesforce sales agent instructing a ServiceNow IT agent to provision hardware while a Workday agent handles employee onboarding.
- Real-Time Enterprise Digital Twins: Artificial intelligence software simulating complete corporate operations in real time, predicting supply chain disruptions, customer churn risks, and cash flow deficits weeks before they materialize.
- Zero-Code Application Generation: Business executives deploying customized, enterprise-grade software applications and database schemas using conversational natural language prompts.
- Sovereign Enterprise Intelligence: Fortune 500 corporations are operating private foundation models behind dedicated corporate firewalls to protect proprietary trade secrets, customer data, and competitive advantages.
Goldman Sachs’ identification of artificial intelligence winners confirms that the technology sector has entered a highly lucrative, execution-focused growth phase. The software companies that master autonomous agent monetization will lead the next century of global economic productivity.
Goldman Sachs’ landmark research identifying the shift in artificial intelligence monetization beyond software seats marks a defining turning point for the global technology industry. By proving that the future of enterprise software lies in consumption-based pricing, autonomous digital agents, and measurable business outcomes, the investment bank has established a clear blueprint for the next phase of the digital revolution. From Salesforce’s $2.00-per-conversation Agentforce benchmark and ServiceNow’s high-margin Pro Plus automation to the massive consumption engines of Microsoft Azure and specialized data gatekeepers like Snowflake and Datadog, the commercialization of artificial intelligence is breaking through historical software ceilings. As the global enterprise software market expands toward $1.0 trillion, the technology leaders that successfully replace manual human labor with autonomous software intelligence will command the heights of the modern global economy.
