💬 [E3233] Salesforce decelerated from steady ~20% organic growth through 2021 to 9% currently and may not have troughed. Adobe fell from 21% in fiscal 2021 to stable 10%. ServiceNow slowed from 29% subscription revenue growth in 2022 to 19% guidance. Multiple compression reflects these fundamentals independent of AI disruption fears.

💬 [E3238] UBS maintains Buy ratings on Microsoft, Snowflake, Datadog, Okta, Zscaler, Oracle, Autodesk, AppFolio, Twilio, Braze, Amplitude, and ServiceNow. Neutral on Adobe, Salesforce, Palantir, and Workday. Smaller-cap preferences favor non-seat/usage-based pricing models and vertical names not at center of AI disruption debate.

💬 [E3213] UBS recommends staying patient on seat-based SaaS stocks rather than buying the dip. They prefer infrastructure and data-exposed names (Microsoft, Snowflake, Datadog) and cybersecurity (Okta, Zscaler) where AI disruption risk is lower and customer spending remains healthy. Among SaaS, ServiceNow and Salesforce could emerge stronger.

🟢 [E3214] UBS's top picks for software exposure are infrastructure/data names (Microsoft, Snowflake, Datadog) and cybersecurity (Okta, Zscaler) where AI disruption risk is lower and customer spending trends remain healthy. These have been 'thrown out with the bath water' in the broad software sell-off. Among SaaS, ServiceNow and Salesforce could emerge stronger. Smaller-cap preferences include usage-based pricing names (Twilio, Braze, Amplitude) and verticals like Autodesk and AppFolio.

🟢 [E3230] The DIY or bespoke option for enterprise AI remains popular and durable. Large enterprises are focused on building custom domain-specific AI agents for back-office and customer support automation. This trend may intensify as OpenAI/Anthropic sell SDKs and hyperscalers lean into DIY-enabling platforms like AWS Bedrock, Azure AI Foundry, and Google Vertex AI.

🟢 [E3217] Microsoft's decision to allocate more GPU resources to its first-party software products signals incumbents gearing up to AI-enable core products aggressively. UBS expects more software firms to invest in re-architecting core products in 2026, potentially declaring 'investment years' with margin pressure to come out stronger.

🟡 [E3204] UBS argues the extreme bear case of outright SaaS displacement is unlikely — even Anthropic, Databricks, and Palantir continue using core systems of record (Oracle/NetSuite, Workday, Salesforce). Incumbents have 10-20 years of embedded workflow knowledge and deep multi-year contract relationships providing reaction time. The 'real' risk is growth moderation and NRR squeeze, not wholesale replacement.

🟢 [E3196] UBS identifies AI-driven change as arriving faster than expected in software, driven by rapid model improvement (Google Gemini 3 Nov 2025, Anthropic Claude 4.5 Dec 2025), growing ability to leverage models into new use cases creating overlap with software firms, declining inference costs, and enterprise AI receptivity evident in Palantir's results. The Street is pricing terminal value risk into seat-based SaaS stocks.

🟢 [E3198] A F500 industrial company using ServiceNow plans 70% AI-deflection of all ticket cases with only L2/L3 handled by humans. They target $200M in productivity gains for 2026, with ServiceNow seats down 30-40% over 2+ years but total spend up high-single to low-double digits due to Now Assist agent usage.

🟢 [E3199] A F50 tech company ML engineer reports using Claude Code from Anthropic to improve software engineer productivity by 35% target and reduce new graduate headcount by 25% unless they are intern returns. Company will have very few positions for new graduate recruiting.

🟢 [E3200] A F500 retail company using Salesforce Agentforce eliminated 80% of outside support seats and 50% of in-house customer support seats since early 2025. The company spent hundreds of millions on data cleanup enabling AI effectiveness. Overall Salesforce spend up 15-20% despite seat reductions due to Agentforce adoption.

🟢 [E3201] A large financial services company using Salesforce targets 15-20% reduction in tier-1 client-facing agents in 2026, representing 400-600 agents. Salesforce spend up 20% due to Agentforce addition, expected to continue increasing in high-single digits versus prior 20-30% growth rates.

🟢 [E3202] AI-native startups have reached approximately $5.3B ARR across 20 disclosed companies, roughly equivalent to AI revenues from public application software firms excluding Microsoft Copilot. This suggests a meaningful portion of incremental AI spend flows to disruptors rather than incumbents, mirroring the SaaS disruption pattern 15-20 years ago.

🟢 [E3203] OpenAI and Anthropic are expected to step more firmly into the enterprise market in 2026 with additional products that compete directly with incumbent software firms. This week's Anthropic plug-in announcements already triggered software stock declines. The risk is that LLM providers sell SDKs enabling enterprise DIY rather than selling to incumbents.

🟢 [E3208] Public application software AI revenues total ~$5.6B with $3.8B from Microsoft Copilot/GitHub Copilot alone. Excluding Microsoft, public apps sector AI revenues are ~$2B. Three years into AI boom, this represents only ~2% of the $290B combined revenue base of the 10 largest public SaaS companies — too small to bend growth curves.

🟢 [E3209] UBS expects software firms to undergo 'AI pivots' in 2026 including: faster re-architecting of core products, lower gross margins from higher AI COGS, shifts to new pricing models away from seat-based, and potential headcount cuts. Street margin estimates showing flat GMs and ~100bps op margin expansion may have downside bias.

🟢 [E3210] Private equity firms hesitate despite software valuation compression because some public software firms are 'unbuyable' due to excessive stock-based compensation expenses. A PE partner notes firms are 'risk off' unable to make sense of AI disruption risk despite credit markets being the best in his career.

🟢 [E3216] A F500 biopharma company had 30% ServiceNow contract increase for AI and other features, with flexibility to reallocate if headcount reduces. Industry expects typical 30% headcount reduction within 2 years, though growth may offset full reduction through repurposing to other parts of organization.

🟡 [E3212] VC partners at UBS summit defended SaaS incumbents, noting 20 years of embedded workflow automation, massive distribution, and valuable corporate data hosting create moats. Multi-year contracts (2-5 years) mitigate near-term disruption, and balance sheets enable protective acquisitions. Consensus: SaaS incumbents not dead, just facing heightened technology change pace.

🟢 [E3211] The legitimate risks getting priced into SaaS stocks are: heightened competition from AI model providers/AI-natives/hyperscalers capturing rents from incremental spend; seat compression with AI agent revenues not offsetting headwind; lower gross margins from AI COGS; and eroding moats as customers share incumbent data with new AI entrants.

🟢 [E3218] UBS's VC Summit in January 2026 revealed heightened uncertainty among VCs heavily invested in pre-AI SaaS firms. While many defended incumbents' moats (20 years of workflow automation, massive distribution, valuable corporate data, multi-year contracts), the consensus was that SaaS incumbents face heightened technology change pace in 2026-2027 putting them in a 'too hard' bucket for many investors.

🟢 [E3197] A F500 hotel company using Salesforce reports physical/human agent seats dropped 10% in 2025 and will drop an additional 30% in 2026 due to Agentforce adoption. The seat reduction cascades to Microsoft licenses, HR system licenses. Net Salesforce spend still increases due to Agentforce and data cloud usage offsetting seat loss.

💬 [E3215] Private equity activity in software has been notably quiet despite valuation compression. A Thoma Bravo partner noted some PE firms are 'risk off' because they can't understand AI disruption risk. Additionally, some public software firms are 'unbuyable' due to excessive stock-based compensation (SBC) expenses, which dilute returns even at compressed valuations. Credit markets are favorable but PE remains cautious on software.

🟢 [E3207] Sharp rotation within tech sector since Sept/Oct 2025 shows software (IGV -21%) dramatically underperforming semiconductors/hardware (SOXX +46%, hardware basket +29%). Consensus view is that AI value accrues to bottom layer of tech stack (processors, servers, storage, AI models) rather than SaaS/application software layers on top.

💬 [E3243] Software sector valuations remain high on strict EPS basis — median non-GAAP P/E of 23x, with many companies not valued on GAAP EPS due to 15%+ stock-based compensation as percentage of revenue. A collective pivot from FCF/revenue multiples to stricter EPS metrics could cause further weakness beyond current correction.

💬 [E3232] Application software sector has decelerated from pre-COVID high-teens/20%+ growth to 12-13% currently, four years into post-COVID slower growth period. This deceleration reflects sector maturity and high penetration rather than AI disruption alone. Growth algorithm has shifted from new logos/seats to price increases and bundling.

🟢 [E3206] The three big software prints from Microsoft, ServiceNow, and SAP last week signaled growth is stable at best but likely still moderating, triggering massive software sector selloff. Growth rates must accelerate or estimates revise upward to bend the negative narrative; more partnerships between SaaS firms and OpenAI/Anthropic would help signal coexistence.

🟢 [E3205] Semis/hardware companies are disclosing AI revenue impacts materially larger than software firms and growing at accelerated clips. Microsoft's Azure acceleration in Q3/March 2025 boosted the 'long-infra' trade. Infrastructure, data, and security software basket up 63% since July 2024 versus SaaS basket down 18%.