At the highly anticipated Goldman Sachs Communacopia + Technology Conference, top executives from leading software companies focused on enterprise and consumer AI applications shared their latest forecasts on future growth prospects. In a fresh note, Goldman analysts highlighted comments from Databricks CEO Ali Ghodsi, who argued that the competitive battleground in AI application software is shifting toward real-world enterprise deployment. Software platforms capable of efficiently and precisely converting model capabilities into dependable business outcomes are poised for new growth opportunities.
Databricks leverages Genie to handle enterprise analytics needs, uses Unity Catalog to unify AI governance, security, and cost management, and expands into customer data platforms, security analytics, observability, and transactional databases. Goldman Sachs emphasized that when assessing the market share growth potential of AI-related application software companies over the next three years, investors should focus on technical intellectual property, innovation velocity, and open architecture. From an investment perspective, Goldman noted that software platforms with vast proprietary datasets and robust workflow advantages are likely to undergo a fresh round of market revaluation.
The analysts also observed that global capital is rotating from the first phase of AI investment鈥攃entered on GPU, HBM, and AI data center infrastructure constraints鈥攖oward second-phase application winners capable of converting tokens into enhanced enterprise productivity, revenue, and cash flow. Valuation divergence may intensify: software firms with exclusive data, complex workflow advantages across enterprise and consumer users, closed-loop agent execution, and clear return on investment are set for revaluation, while traditional SaaS models vulnerable to commoditization by foundational models could face continued pressure.
AI Delivers Results: Enterprise Data and Governance Determine Deployment Success
At the Communacopia + Technology Conference, Genie illustrated Databricks鈥?strategy for capturing the enterprise AI entry point. Through proprietary data indexing technology OntoRank and business ontologies, Genie organizes enterprise data into business information that models can understand, compute, and utilize, positioning the product as an "AI agent analyst." Databricks management indicated that Genie is widely deployed across internal sales, marketing, and finance functions, and has significantly transformed operations over the past 6鈥?2 months.
The engineering significance lies in the fact that enterprise agent deployment requires unified understanding of business objects such as customers, orders, and revenue, with analysis results integrated into daily workflows. Even as model performance improves, these data foundation tasks remain critical to task accuracy. Management also sees collaboration potential with Palantir on complex data engineering projects. Unity Catalog addresses governance needs arising from large-scale agent deployment.
The Goldman analyst team pointed out that rapid model iteration increases selection and deployment complexity, token consumption growth intensifies scrutiny over costs and return on investment, and enterprises still lack sufficient visibility into agent operations, data access, and security risks. Consequently, data governance, access permissions, security, and business semantics must be tightly integrated with AI runtime environments. Goldman鈥檚 latest assessment of the AI application layer is that such platforms deliver commercial value by enabling enterprises to scale deployment confidently鈥攃larifying what agents can access and execute while tracking task costs and outcomes.
Customer stickiness increasingly derives from sustained business value creation, aligning with Goldman鈥檚 emphasis on technical capability, innovation speed, and open architecture. Platform leaders like Databricks are expanding product offerings across four directions. In customer data platforms, Infinite Campaigns supports a shift from segment-based marketing to individualized precision marketing, with management anticipating that open-source models will further enhance cost feasibility. In security information and event management (SIEM), new data generated by agents, combined with the convergence of CIO and CISO responsibilities, creates opportunities for data platforms to enter security analytics. Observability is another expansion area, though no product timeline has been disclosed yet.
Lakebase is viewed by management as potentially the largest new opportunity: as AI lowers software development barriers, new applications require transactional databases to support daily read-write operations and business functioning, thereby expanding Databricks鈥?market beyond analytics. Management even suggested that software created over the next eight months could surpass cumulative historical output鈥攁lthough Goldman noted this is an aggressive demand projection, not a validated industry forecast.
Turning Tokens Into Cash Flow: Application Opportunities Broaden, Software Stocks Diverge Rapidly
OpenAI鈥檚 major Astra launch focuses on agent-driven efficiency gains in computer operations, coding, and complex task execution, making more enterprise workflows automatable. However, "model performance competition" and "business deployment competition" will advance in tandem. Pricing models are also becoming tiered: Astra鈥檚 API still charges per input and output token, while AI application software vendors can charge per seat, operation, conversation, or business outcome.
For example, Intercom鈥檚 Fin charges per defined business outcome, starting at $0.99 per outcome, while Salesforce鈥檚 Agentforce offers billing options per operation, conversation, and user license. Foundational model providers will likely continue focusing on selling computational resources via APIs, while upper-layer AI applications must increasingly prove delivered value. Outcome-based pricing aligns prices more closely with customer benefits but requires vendors to absorb costs related to inference, retries, and human review. Task success rates and profit per task therefore become key commercialization quality indicators.
Recent stock market movements already show the market rewarding realized AI revenue while penalizing potential substitution risks. On September 3, Snowflake, a Databricks competitor, surged roughly 17% following strong earnings. Its second-quarter product revenue grew 37% year-over-year to $1.49 billion, with full-year product revenue guidance raised from $5.84 billion to $6.07 billion, providing operational evidence that the enterprise AI application layer is expanding data platform consumption.
By September 8, however, Salesforce and Intuit shares each fell approximately 4%, ServiceNow dropped about 5%, and the S&P 500 Software & Services Index declined 1.4%. Wall Street analysts attributed part of the selling pressure to Astra鈥檚 emergence, which once again triggered concerns about large language models replacing software.
Databricks鈥?software deployment path underscores how AI gains are progressively spreading across the industry chain. As agents execute more tasks, they drive consumption of compute, storage, and data services. Platforms with enterprise data connectivity, governance capabilities, and business process advantages are competing to convert this usage into revenue and cash flow. For software stocks, this constitutes a structural tailwind, but platform expansion will also squeeze market space for certain standalone tools.
Companies positioned to benefit must simultaneously deliver on new AI revenue, expanded customer consumption, and profit margins. Vendors that rely on interfaces or single-function advantages will face intensifying competition.
The valuation anchor for the AI investment wave that began in late 2022 is progressively shifting from "capital expenditure scale" to "capital return efficiency," and this process is accelerating. In other words, while the first phase of the AI investment frenzy focused entirely on "who controls and profits from building the largest GPU data centers," the current second phase centers on "who can convert tokens into sustainable cash flow."
The AI-driven bull market is moving from "buying chip stocks" toward "buying AI workflows." The market is repricing the primary investment theme from "who benefits from rising AI capital expenditure" to "who can fastest convert computing power into ARR, margins, and free cash flow." This latest rotation favors AI application platform companies embedded in critical enterprise processes, with high renewal rates, data moats, and agent monetization capabilities. Whether AI-focused software firms can continuously access stronger, more cost-effective models while preserving their irreplaceable business value will determine their long-term pricing power.