"Software is dead, AI should rise"?

The AI wave is reshaping the technology industry with unprecedented momentum, sparking intense debates about the future of traditional software businesses.

In the past two weeks, sentiment in the software industry has clearly turned bearish. After OpenAI released its latest large model GPT-5 last Tuesday, concerns about AI replacing traditional software triggered panic selling in the market, causing the stock price of German software giant SAP to plummet by 7.1% at one point, wiping out nearly €22 billion in market value, marking the largest single-day decline since the end of 2020, and dragging down European software stocks collectively.

According to reports from the Chase Wind Trading Desk, Goldman Sachs analyst Gabriela Borges and her team stated in a report released on the 17th that investors' concerns are centered around a core survival risk: Will AI become a disruptive force that erodes existing pricing models, lowers the barriers for new entrants, and ultimately compresses the profit margins of SaaS (Software as a Service) giants?

In Goldman Sachs' view, the answer is no.

The report pointed out that the notion that "software is dead" is overly pessimistic. In some cases, AI has the potential to be a "force multiplier" for industry-leading suppliers. The current phase is similar to the historical transition of the software industry from on-premises deployment to cloud computing, which gave rise to a new batch of leaders while also prompting established companies like Adobe, Intuit, and Autodesk to transform into larger, faster-growing, and more profitable enterprises, creating significant excess returns for investors.

Goldman Sachs expects that as the pressure from enterprise software renewal cycles eases in 2026, the new contributions brought by AI will provide stability for key metrics such as Net Revenue Retention (NRR), paving a lasting and multi-year growth path for the industry. The future leaders of the software market will be composed of two types of companies: the current giants that continue to innovate and the emerging companies that successfully build differentiated AI-native software.

Is AI a Disruptor or a Force Multiplier?

In the debate over whether AI-native companies can replace traditional SaaS companies, the core issue is whether their products can be "significantly better and cheaper."

Goldman Sachs pointed out that today's SaaS giants are not comparable to the on-premises software companies they once disrupted. At that time, SaaS had a significant advantage in cost and experience due to its cloud architecture, whereas today the competitive barriers set by SaaS companies are quite high.

In terms of pricing strategy, the biggest risk faced by AI-native companies is providing a value-based pricing model, which could pose an erosion risk to seat-based models. However, SaaS leaders have begun to evolve their pricing models towards outcome-based methods, such as Salesforce's approach, or incorporating AI features into existing subscription agreements to encourage adoption, like HubSpot.

In terms of functional innovation, SaaS leaders maintain a rapid pace of innovation through organic growth and acquisitions. Cases such as ServiceNow acquiring Moveworks, Braze acquiring Offerfit, and Salesforce acquiring Bluebirds are numerous In terms of organic innovation, Salesforce will launch Agentforce in September 2024, which largely defines the next stage of AI adoption.

As a result, the report believes that it is very difficult for AI-native companies to disrupt existing SaaS leaders in terms of pricing and product features.

A case study by Goldman Sachs shows that in call center scenarios, AI applications can reduce operational costs for businesses by 30%, but the total software budget actually increases by 130%. This indicates that AI not only does not compress the market but may actually expand the total addressable market (TAM) for application software.

Hybrid AI Strategies and Moats of Software Giants

To address challenges and seize opportunities, giants in the enterprise software sector are widely deploying a hybrid AI model strategy, which has become key to solidifying their moats.

This strategy combines domain-specific models trained on proprietary data with external cutting-edge large language models (LLMs), retaining their data advantages while providing flexibility and high performance.

From Snowflake's Arctic model, Salesforce's Agentforce to ServiceNow's Now LLM, these industry leaders are embedding their proprietary models to handle core enterprise workflows while allowing customers to interface with external models like OpenAI and Google.

The report explains that this strategy greatly reduces the risk of being undermined by AI-native newcomers, as it firmly locks customers into a familiar, secure, and deeply integrated ecosystem. Meanwhile, vertical software vendors have stronger barriers due to their profound industry knowledge.

Goldman Sachs believes that in these specialized fields, AI is more likely to play an "augmenting" rather than "replacing" role. Due to the complexity of workflows and high costs of data migration, even if AI-native product technologies are outstanding, it is difficult to gain customer trust and adoption in the short term.

Higher Barriers for Enterprise Software Compared to Consumer Software

The report also points out that the market often underestimates the fundamental differences between enterprise software and consumer software, which constitute another important moat for existing giants. Goldman Sachs emphasizes that the entry barriers for enterprise software are much higher than for consumer applications, with the core being "mission-criticality."

The report notes that the "hallucinations" of AI models may be harmless in consumer scenarios, but can lead to serious consequences such as reputational damage and customer loss in enterprise environments. Therefore, enterprise software is not only about product functionality but also involves a complex system "beneath the iceberg."

Additionally, there is a current view that customers can build their own SaaS equivalent tech stacks at lower costs than ever before. Agent AI significantly lowers development barriers through code generation and debugging tools, and reduces the barriers to complete UI through natural language-driven agents. For example, Palantir's value proposition combines cutting-edge deployment engineers, ontology operating systems, and AI toolkits However, Goldman Sachs believes this boils down to the classic issue of insourcing versus outsourcing. Even if the cost of maintaining software applications internally decreases, the costs of professional vendors will also decline, so the cost-effectiveness frontier of professional vendors typically always leads that of internal technology.

Goldman Sachs argues that simply replicating a software stack with AI is not only fraught with difficulties but also "somewhat meaningless." The essence of technological change is to create new, differentiated applications rather than to reshape the past.

The report points out that the industry may currently be at a local peak in the opportunity share for customized versus packaged solutions: existing SaaS companies lack comprehensive AI capabilities, and AI-native companies typically only address a small part of broader enterprise issues. Many developers and IT professionals are still in the early stages of the learning curve regarding the successful implementation of AI projects.

Future Indicators to Watch

Despite maintaining an optimistic long-term outlook for the software industry, Goldman Sachs also highlighted several key factors that investors need to closely monitor to assess whether the arguments of both bulls and bears are validated.

First is the stability of Net Revenue Retention (NRR). The report shows that over the past two years, SaaS leaders have faced significant renewal pressure, as pandemic-related demand peaked in the first half of 2021 and 2022, and these three-year contracts will face the greatest pressure upon renewal in the first half of 2024 and 2025.

Second is the contribution of AI revenue to growth. SaaS leaders need to demonstrate that their AI products can generate incremental revenue. For example, Adobe expects its AI products to contribute $250 million in annual recurring revenue (ARR) by the end of 2025, and the report notes that this clear financial data will be a key validation signal.

Third is customer feedback on the innovations of SaaS leaders. The market will focus on the progress SaaS leaders make in addressing AI application barriers (such as model selection and tool integration).

Finally, the development momentum of AI-native companies will be assessed to determine their durability and the risks they pose to existing enterprises. Many AI-native companies established over the past two years have the potential to evolve into larger-scale enterprises over time, expanding from single-product companies to suite products, and long-term tracking of such companies will help evaluate their actual impact on existing profit pools