美国中小盘软件：与Thoma Bravo的AJ Rohde一起在AI领域获胜

Software: Winning in AI with AJ Rohde from Thoma Bravo Last night we were fortunate to catch up with AJ Rohde, Sr. Partner at Thoma Bravo,along with an all-star cast of software investors for discussion and dinner. The top-of-mind topic is probably easy to guess: AI disruption. And then more AI — real-world use,margin improvement, and impact on portfolio pricing. Of course, we couldn’t get by without Peter Weed+1 917 344 8390peter.weed@bernsteinsg.com Luwei Yang+1 917 344 8342luwei.yang@bernsteinsg.com Overall Thoma is optimistic about AI’s contribution to returns.Their strategy seemsto be well-positioned vs. AI disruption — a portfolio focus on cybersecurity and enterprisevertical-applications with strong data gravity (and low churn) both appear prescient, andremain areas where they think continued focus is wise. These categories succeed dueto strong customer ownership and the challenge new AI-native entrants face to break in.Not only are enterprise customers slow to move, but AI-coding tools allow the portfolio to Armin Hadavi, CFA+1 917 344 8463armin.hadavi@bernsteinsg.com Limited seat-based pricing may position their portfolio well for value capture.Justbefore COVID they asked the 72% of portfolio companies using seat-based prices tofind a better-aligned meter. Success: now only 21% are left seat-based, demonstratingtransformation is possible. For instance, some companies moved to more ROI-based The biggest PE investment challenge? Not targets or valuation… it's liquidity.Theythought targets exist and valuations are actually pretty attractive relative to history, but theliquidity environment is really tough. They frequently struggle with confidence on an exitscenario to IPO or strategic buyer vs. history (and selling to other PE is not a great way todrive returns). Recent IPOs prices have been weak, and they perceive strategics are feeling Software leaders and businesses that impress them most? A few surprises!The question arose on which software vendors are strongest, and where they shape theirportfolio to avoid direct competition. Microsoft, AWS, and ServiceNow platforms came upfirst — they see them as clear winners with dominant control of their customers and rollingup their spaces. And in cybersecurity, they add Palo Alto and CrowdStrike to the list. Theyrespect leaders of these businesses deeply (Bill from ServiceNow and Nikesh from PaloAlto spend a lot of time with them), but beyond those, AJ pointed immediately to Atlassian as Finally, the topic of AI’s operational use and bottom-line impact was of deep interest.In service desks and back office functions, they think an incremental 50%improvement is possible. COGS is the other opportunity, particularly on support costsand better vendor negotiation. On the other hand, they don’t see sales as a cost-takeoutopportunity. In R&D, they’d like more leverage from reducing effort in maintenance /refactoring of legacy code (it can be the majority of R&D team time). And rotate more R&D BERNSTEIN TICKER TABLE INVESTMENT IMPLICATIONS No change to our ratings or price targets. DETAILS Key discussion topics and insights of the night: •Markets facing “low-hanging” AI disruption risk: service desks •Markets he is more optimistic on: the office of CFO software, cybersecurity, critical infrastructure software, industrial designsoftware, and vendors with their own proprietary data, like the information services companies Incumbents with their deep customer relationships and data gravity will continue to hold competitive advantages over new •Revenue deceleration prompts more focus on margin expansion with cost savings driven by AI •But some areas will receive big investment focus: sales engineer, product teams, channel ecosystem, and partnerships •Companies have been shifting away from a seat-based pricing model to using other metrics that better reflect ROI •Palo Alto’s acquisition of CyberArk has a strong strategic fit A main topic of discussion was around AI disruption, and which markets would face a higher risk.One area wherewe are already seeing AI disruption are in service desks broadly. For instance, in customer support, most of the L1 customersupport, which are the basic issues that can be easily scripted like resetting passwords, have already been automated. With theadvancements in GenAI, many of the L2 customer support tasks, which are complex problems that require more expertise, canalso be automated. It’s only for the most complicated issues in L3 support, such as software bugs and data recovery, that AI hasnot been able to automate and still requires direct human expert involvement. Another market with a high risk of AI disruption On the positive side, some areas where AJ is more structurally optimistic on are:office of CFO softwaresuch as generalledger applications;cybersecurityand more specifically identity management, operational technology, and industrial securitysolutions, but need to be areas with limited competition against