Please refer to our disclaimers, which can be found in the footnote of this page and here.

Contents

• Contents
• Performance
• Portfolio Discussion
• Software Disruption Follow Up
• Guidewire
• Business Overview
• Core Products
• New Products and AI
• Data and Analytics
• Business Quality & Cost of Capital
• Why do we Own It?
• Mission-Critical Software
• Market Leadership
• Deep Structural Moats
• Long-Duration Growth and Cash Flows
• Key Risks
• AI Execution and Disruption
• Leadership and Equity Alignment
• Legacy Codebase and Technical Debt
• Valuation[1]
• Business Update
• Footnotes

Performance

At the time of writing (4 Nov 2026), year-to-date, the portfolio is up +8.1% gross / +6.7% net (vs the MSCI ACWI at +12.3%). Longer-term, since inception (1 Jan 2019), the portfolio has compounded at +17.0% gross / +14.4% net (vs. the MSCI ACWI at +14.1%), representing +2.9% gross (+0.3% net) annualised outperformance. 3Q 2026 performance was +17.1% gross / +16.9% net (vs. the MSCI ACWI at +1.6%). See table for quarter-end figures.

The top contributors year-to-date are Okta, Guidewire, Microsoft, and Alphabet. The top detractors are Charter Communications and Pinduoduo.

We started the quarter with 1% cash and ended with 11%. During the quarter, we exited our investment in Okta. Okta’s share price has risen materially this year as fears about AI-driven software disruption softened and the market began to price in Okta’s own agentic identity opportunity. We like Okta’s core businesses, but it is too early to know whether Okta will win in agentic identity. As the price continued to rise, meeting our required returns would have required the agentic opportunity to be part of our base case. We could not underwrite that, so we exited.

In 2Q 2026, we made Guidewire a core position at around $102. The share price then rose sharply without any change in underlying value, so we reduced the position at around $206. We outline our thesis on Guidewire below.

During the quarter, CCC Intelligent Solutions received a takeover bid from GTCR and Elliott. Under the offer, investors will receive $7.75 per share in cash. While this is a premium to the pre-announcement share price and a significant markup from the lows when we were deploying capital, it undervalues the business by about 20%. As such, we are disappointed that the board accepted the deal. Given the company’s private-equity history, we knew this was a risk, but we took some comfort in the CEO’s large share ownership. Unfortunately, this did not create alignment for us. The CEO will roll his stake into the private entity, and we assume future management incentives will be easier to achieve at a lower acquisition price. We would like shareholders to vote the deal down and GTCR and Elliott to return with a price more palatable to minority shareholders. Other large shareholders we have spoken with expressed a similar sentiment, and we have also provided this feedback to the management team.

Portfolio Discussion

Software Disruption Follow Up

We presented our thoughts on Software AI Disruption Risk in our last letter. Since then, we have been speaking with software engineering experts to deepen our understanding of technical areas that might create advantages or disadvantages for incumbents or new entrants. Our focus on this continued work ignored the business model challenges and instead looked only at the technical challenges. We wanted to know if there were any software engineering-related reasons that would make it difficult for incumbents to compete, akin to the on-premise vendors during the public cloud transition.

In short, while the AI transition is much more transformational to society and as a technology than was the public cloud, the technical differences between incumbent SaaS vendors and new AI-native-software entrants seem less extreme (note, we are talking about software new entrants here, not the frontier labs).

The move from on-premise software to cloud SaaS changed several things simultaneously. There was a physical infrastructure change, with the vendor taking responsibility for running the servers. The vendor also took over software maintenance and updates (i.e. software went from a product to a service). Upgrade cycles, maintenance, availability, security and compliance also became continuous responsibilities for the vendor. For many applications, there was also a shift from single-tenant to multi-tenant architecture, where customers shared infrastructure. Software designed around one customer’s installation had to work safely and efficiently across many customers. Simply hosting the old product in the cloud did not achieve this, and therefore incumbent on-premise businesses also suffered a deep technical disadvantage.

By contrast, an established SaaS vendor moving towards agentic AI need not undergo a comparable wholesale architectural conversion. It already has many of the foundations that AI-native entrants still need to build. Both incumbents and new entrants must add agentic layers, such as a centralised data/reasoning engine, an ontology layer that maps data and workflows (i.e. context), agentic reasoning logic, connections to deterministic process automation tools, MCP access, and agentic governance. These additions remain significant undertakings.

In various areas, the incumbents have a credible advantage since they own or control the data, context, and client relationships. Good context is a bottleneck. Even in a headless model, the incumbent isn’t reduced to a dumb database; it remains a critical system of record and context layer. Further, for various applications, integrating the agents with the system of record reduces the number of steps between reasoning about an action and taking it, with lower latency and token consumption as well as higher accuracy. A native agent may also have access to context and capabilities that are not exposed externally, allowing more actions to be authorised and more workflows to be automated within defined controls.

That said, incumbents still face real architectural disadvantages. Older SaaS incumbents often have monolithic software architectures (everything in one big codebase) versus more modern microservices stacks. Building, testing, and releasing changes across a monolithic core takes time, although the controlled release process also protects reliability. Incumbents must preserve existing APIs, customer customisations and automation, which complicates agent development and testing. A new entrant can design a narrower product around agents from the start and change it more quickly. Further, as mentioned in our last letter, re-tooling old human workflows for agentic execution may be harder and take longer than simply building new workflows from scratch. Finally, while incumbents have a context advantage, they still do not capture all the context in customer decision-making processes. Much of this lies in things like emails, Slack channels, and in people’s heads. Incumbents and new entrants need to capture this too.

Overall, our further work on the technical areas led to the same view presented in our prior letter. Mission-critical, enterprise software vendors with credible AI roadmaps have significant potential in the AI era.

Guidewire

Business Overview

We just came back from Guidewire Connections, the company’s annual customer conference in Las Vegas. We joined sessions on recent product releases and the roadmap, attended an investor session with management, and spoke with Guidewire employees, systems integrators, partners, customers and other investors.

Guidewire provides the operating system for property and casualty (”P&C”) insurers. Its software is the system of record for policy administration, claims and billing, with revenue across the three split roughly 40%, 40% and 20%, respectively. Newer products will extend its reach into pricing and underwriting. Choosing a vendor for these roles is a highly strategic decision for insurers, and vendor trust is incredibly important. The decision usually goes to the board. A large insurer can spend years migrating state by state and line by line. Contracts can run for as long as 10 years.

We first underwrote Guidewire in 2020. Our assumptions were too conservative, and we passed. Since then, we have spoken extensively with former employees, partners, industry experts and customers. The stock entered our range in 2022, but the broader sell-off presented other attractive opportunities that held our attention. It subsequently moved beyond our target range, and by 2025 we considered it significantly overvalued. The market was extrapolating cloud growth that was one-time in nature (on a per customer basis) as if it were perpetual.

We continued to follow Guidewire and bought during the Q1 software sell-off. We increased the position materially when the shares hit $102 in Q2. Two months later, the shares had more than doubled, with no change in our assessment of fundamental value. We still found the valuation attractive, but the position had grown too large for the risk-reward, so we trimmed.

Founded in 2001, Guidewire began by selling on-premises software. Its cloud transition began in earnest in 2016 and accelerated in 2019 when Mike Rosenbaum joined as CEO. Both the company’s history and its customers’ conservatism contributed to the late start. We estimate that the cloud now accounts for 60% or more of the direct written premiums (DWP) Guidewire serves and 84% of its annual recurring revenue (ARR). We expect those shares to reach 90% and 95%, respectively, over the next decade.

Roughly 2,400 P&C insurers write about $3 trillion in annual premiums globally. Personal lines account for just over half, with commercial lines making up the rest. Just 340 insurers underwrite more than 85% of global DWP, and 90 underwrite 66%. We expect DWP to grow in line with nominal GDP because premiums track the nominal value of insured assets, which in turn tracks the global economy. That link, combined with Guidewire’s pricing model, gives the business a natural long-term inflation hedge.

Just under 30% of global DWP runs through Guidewire in some form, making it the clear market leader. Given the strength of its product and its record of winning business, we expect this percentage to rise.

Vendors in this market typically charge a percentage of the DWP that runs through their platforms. Guidewire earns about 14 basis points on roughly $860 billion of DWP, or 19 basis points on a fully ramped basis[2]. We expect the rate to rise over the coming years as customers migrate to the cloud at ~2.5 times on-premises pricing and adopt more products. The fee is a small share of an insurer’s costs, though a larger share of its profits. In the example below, a fee of 20 basis points of DWP amounts to 21 basis points of costs and 4% of profits.

In FY26, Guidewire generated $1.5 billion of revenue. ARR was $1.25 billion, or $1.6 billion on a fully ramped basis. Gross margin was 64% overall and 73% on subscription revenue. Operating margin was 10%, and free cash flow (FCF) margin was 24%, or 12% after deducting stock-based compensation (SBC).

Core Products

Guidewire’s core offerings are InsuranceSuite, its policy, claims and billing applications, and InsuranceNow, a simpler integrated platform for smaller insurers. InsuranceSuite accounts for 90% of cloud ARR.

InsuranceSuite

• PolicyCenter manages the full policy lifecycle, including product configuration, rating, quoting, binding, issuance, endorsements, audits, cancellations and renewals.
• ClaimCenter manages claims from first notice of loss (FNOL) through investigation, assessment, reserving, settlement, payments, litigation and recoveries. It automates workflows and integrates with third-party service providers.
• BillingCenter manages premium billing and receivables, including invoicing, collections, instalment plans, agent commissions, delinquencies and reconciliation. It integrates with external payment providers.

InsuranceNow combines policy administration, claims, billing, document management and analytics in one cloud platform. It primarily serves smaller and mid-sized P&C insurers and managing general agents (”MGAs”) that need a more standardised, less complex offering than InsuranceSuite.

New Products and AI

In our two prior letters, we outlined the characteristics of software companies we are looking for. A key part of that is execution in AI product offerings and the future roadmap. Guidewire appeared to lag some of our other portfolio companies in this aspect. However, we left the conference impressed by both the product suite and the roadmap. Many of the products had launched only weeks earlier. Management and employees spoke with a sense of urgency about this work. Customers expressed excitement but noted it was early days. Continued customer feedback and adoption will be central to our ongoing diligence.

Some examples of Guidewire’s AI products include:

• ProNavigator is a generative AI assistant embedded in Guidewire’s core applications. Insurers can use it to retrieve information, summarise claims, answer questions and automate selected workflows.
• Concierge automates first notice of loss through conversational AI agents. The agents collect information by voice or chat and create claims directly in ClaimCenter.
• UnderwritingCenter uses AI to automate submission intake, extract information, evaluate risks and support underwriting decisions.
• PricingCenter uses advanced analytics and optimisation to help insurers develop, evaluate and deploy pricing strategies.
• Agentic Framework and Agent Studio give insurers the infrastructure and tools to build, deploy and govern autonomous AI agents that interact directly with Guidewire’s core systems and carry out insurance workflows.
• Guidewire also bought hyperexponential (”hx”), which we discuss below.

Data and Analytics

Guidewire’s data management, business intelligence and predictive analytics products help insurers bring together, analyse and act on policy, claims, billing and external risk data. Data Platform, Data Studio, Explore, DataHub and InfoCenter handle data integration and reporting. Predict, Industry Intel and Compare cover predictive modelling and industry benchmarking. HazardHub, Canvas and Cyence provide specialised property, catastrophe and cyber risk analytics. We see these products as useful extras, but have limited conviction in their advantage over systems insurers can build with Snowflake, Databricks or other internally managed data systems.

Business Quality & Cost of Capital

Why do we Own It?

Mission-Critical Software

An insurer’s policy and claims software is its operating system, its primary system of record and the place where much of the work gets done. A claims manager, for example, might spend 80% or more of their working day in it. It integrates with every other major IT provider the insurers use. Insurers have little reason to cut costs here.

The importance of these systems and the cost of switching make replacement a strategic decision that carries significant risk. Boards take a view measured in decades. A CIO might oversee one such transition in a career, or two at most. Recommending a vendor outside the major providers is politically risky for the CIO. Guidewire is the safest choice.

Every customer we spoke with had ideas about what Guidewire could do better. Yet each spoke about the future as though continuing to use Guidewire were a given.

Market Leadership

Guidewire’s lead has strengthened since we first underwrote the business in 2020. At the time, moving to the cloud was the main focus for both Guidewire and its customers. A migration was already a major undertaking, so it gave customers a natural opportunity to reconsider their vendor. Guidewire executed well, and it became easier for prospective customers to find peer references with each migration.

Duck Creek is Guidewire’s most significant competitor. It also began on-premise, but its smaller size allowed it to start its cloud transition earlier. In 2020, customers thought more highly of Duck Creek’s cloud offering. Our view at that time was that Guidewire’s structural advantages would allow it to compete effectively despite the later start. Customer perceptions have since reversed. Guidewire is now seen as the leader in both on-premise and the cloud.

Vista Equity Partners took Duck Creek private in 2023. Our best estimate is that Duck Creek serves $150 billion or more in DWP against Guidewire’s $860 billion. Guidewire serves about five times as much DWP globally. Duck Creek is more concentrated in the US, but even there Guidewire serves about three times as much DWP and likely generates around twice the revenue.

Guidewire’s win rates have been consistently strong for several years. They are particularly high when weighted by DWP because Guidewire does better with larger insurers. For large US insurers, the market is effectively a duopoly between Guidewire and Duck Creek. Both face more competition for smaller insurers.

Deep Structural Moats

Software moats typically come from switching costs and, in some cases, a sales and distribution advantage built on existing customer relationships. Guidewire’s switching costs are among the highest we have seen in software. Customers tell us that even migrating to the cloud with the same provider can take about 18 months. Replacing a vendor takes years, sometimes many, as insurers move state-by-state and line-by-line.

Given its “operating system” status, rebuilding its integrations accounts for much of the work and cost. Total costs can exceed $100 million for a Tier 1 carrier and run into the tens of millions for a mid-sized carrier. The business must keep operating throughout the switch. It is a “rebuild the aircraft while flying” moment. Historically, customers have made substantial customisations, often to about 1% of the codebase, though Guidewire is working to standardise these in its cloud deployments.

Historically, overall churn has ranged from 1.5% to 3% and has trended lower. Churn for core systems has been below 1%. Apart from mergers, Guidewire has not lost a customer with more than $1 million of ARR to a competitor in the past five years.

Guidewire’s excellent relationships with insurance executives give it a strong sales and distribution advantage, reinforced by its network of systems integrators and partners. We saw this at the conference, which integrators and partners consider a key opportunity to meet insurance customers. Customers also spoke about how much they trust Guidewire.

Long-Duration Growth and Cash Flows

We think investors often underestimate the duration of Guidewire’s growth and how sustainable its cash flows are. Many companies we assess need new markets or adjacent products to sustain growth. Those opportunities typically carry more investment risk than expansion in the core business. Guidewire can grow for years in its existing markets alone. As we discuss in our Business Quality Framework, that matters to our assessment of business quality.

Guidewire currently has $1.2 billion in ARR, or $1.6 billion fully ramped. The company sees an additional $4 billion of opportunity from cloud upgrades and deeper penetration of existing customers, plus $8 billion from new customers. These figures exclude other products such as analytics and newer agentic products. Guidewire says agentic products could add another $75 billion to its total addressable market. However, our base case includes only the core opportunity and treats agentic expansion purely as upside.

Of the $3 trillion in global DWP, Guidewire already serves $860 billion in some capacity. Its record of winning business suggests that both its share of DWP and the number of products each customer uses will continue to rise.

From customer conversations and industry data, we estimate that 30% to 35% of the market still uses systems built in-house. About half of those systems are outdated. Given the advantages modern vendors offer, we expect much of that business to move to external providers over the next decade, with Guidewire being a key beneficiary. One very large insurer told us it had spent more than a decade building its own claims management system before concluding that it was wholly inadequate and that it could never keep pace with the best third-party technology.

Guidewire will also have opportunities to replace other vendors, although its competitors benefit from high switching costs too. We think the most likely openings will come when insurers merge, decide whether to move to the cloud or replace a system in a particular country or region.

We have high confidence in the revenue from Guidewire’s own cloud transition and see it as a lower-risk source of growth — Guidewire Cloud is the natural choice for an insurer migrating to the cloud from Guidewire on-premise. We estimate that the cloud accounts for 60% or more of the DWP Guidewire serves and 84% of ARR, rising to ~90% and ~95%, respectively, over the next decade. Each migration produces a one-off step-up in pricing for that customer. Migrations have accelerated revenue growth in recent years and, as mentioned above, the market has, at times, extrapolated that pace and overvalued the shares.

Our base case assumes compound annual growth of 14% in revenue and 18% in subscription revenue over the next decade. Without the benefit of pricing gains from increased product penetration with existing customers (i.e. surplus to any cloud uplift), Guidewire would need to grow global DWP penetration from 30% to 55% to reach these CAGR levels. At just 2% incremental pricing gains per year, it would need to reach just under 50%. To note, DWP penetration differs from market share because two vendors can serve the same premiums in different capacities. Any pricing gains from additional products would reduce the DWP penetration required (see below chart). We think Guidewire’s market position and win rates support a pricing and penetration combination that can meet or exceed this growth rate over the next decade.

Our confidence in the durability of Guidewire’s cash flows rests on how essential its products are to insurers’ operations and how strategically insurers choose these systems. Customers spoke as though Guidewire would certainly be part of their future, at least in the role it plays today.

Key Risks

AI Execution and Disruption

We see AI disruption, poor execution of the AI roadmap, or both as the principal risk and the one weighing on the share price. In our prior letter, we discussed our research into how customers view AI and their strategic software vendors. Our view differs from the market’s. We expect vendors of mission-critical, enterprise-grade software to do well in the AI era if they have strong AI roadmaps and deliver on them.

The progress we saw at the conference eased some of our concern about Guidewire’s slow start in AI. Management and staff conveyed the same urgency about AI in formal sessions and informal conversations.

It is still early for Guidewire’s AI products. The company has announced deal wins but has yet to disclose AI ARR or unit economics. Revenue from AI products is likely minimal today. Customers told us they were enthusiastic and wanted to trial the products. Whether that interest translates into adoption remains an important question for our diligence.

Guidewire accelerated its roadmap with the recently announced acquisition of hyperexponential (hx), an agentic pricing and underwriting platform for commercial and specialty insurance. Guidewire’s CEO described the deal as the most strategic decision of his tenure. It raised questions for us about the performance of Guidewire’s own PricingCenter and UnderwritingCenter. When pressed, management said hx mostly serves different lines of insurance, with some overlap.

We recognise the strategic rationale for the deal, but the price raises concerns about financial discipline. Guidewire paid $1.7 billion in cash and debt, with another $400 million payable if hx meets certain performance targets. In FY27, hx is expected to generate about $80 million in ARR in FY2027 and we estimate revenue of about $65 million. The business breaks even on free cash flow. If hx meets the targets, the price equates to 26x ARR or 32x forward revenue.

These are euphoric multiples. On our rough assumptions[3], hx would need to grow revenue at a 34% CAGR over the next decade for Guidewire to earn a 15% unlevered IRR. In our view, the price makes sense only if the acquisition also removes an existential threat to the core business. Leverage improves the return, but given the venture-like risk, we assess the deal on an unlevered basis.

The price looks even more generous from hx’s perspective. To grow at that rate, hx will likely need exceptional sales and distribution, and we think Guidewire is better placed than anyone to provide this for them. The price appears to give Guidewire no credit for that contribution. hx got a good deal.

Leadership and Equity Alignment

The CEO owns $28 million of stock and earns around $15 million a year. He sold about $50 million of stock over the past three years, though only half came from voluntary sales under a prearranged trading plan. The rest was sales to meet SBC tax payments. His required shareholding is only four times base salary, and base salary is a small part of his compensation.

Stock awards account for the vast majority of his compensation. Half of the awards are time-based and half are performance-based. ARR growth carries an 80% weight in the performance component, and adjusted operating income less SBC carries 20%. Both measures are subject to discretionary adjustments for acquisitions.

We have two concerns about these incentives. First, suppose Guidewire misses its ARR growth target and the shares fall 30%. The CEO would lose around $8 million on his holding, but the part of his pay that is independent of performance would roughly offset that loss. We want management to share in shareholders’ losses and think the ownership requirement should be much higher.

Second, although Guidewire can adjust performance targets for acquisitions, the absence of a total shareholder return (TSR) or return on invested capital (ROIC) measure creates an incentive to buy faster-growing businesses to boost future ARR growth, with less regard for the price paid. Against that backdrop, the high price paid for hx is less surprising.

Legacy Codebase and Technical Debt

Our work on Guidewire’s technology stack included conversations with former members of its technical staff. Its core platform was built for on-premises deployment. Before the cloud transition, it comprised around seven million lines of code, and a typical customer might have customised about 1% of the codebase. PolicyCenter, ClaimCenter and BillingCenter share components and come with substantial prebuilt integrations, but each is a separate application with its own databases. For example, PolicyCenter and ClaimCenter each maintain customer records even when both serve the same customer.

Rebuilding the platform from scratch would have been a monumental undertaking. Guidewire essentially moved each customer’s core applications to the cloud using a single-tenant architecture. Customers run their own application instances in isolated virtual environments within a shared Customer Cluster. Some instances retain customer-specific customisations. This design is inefficient from a margin perspective, but it suits insurers that often require single tenancy for data protection and regulatory purposes. Guidewire has built multi-tenant services around the core and connects them via APIs.

Running a separate application instance for each customer is inefficient. It can also cost more to develop functionality embedded deep in the core or to scale individual components. The separation of the core applications also makes it harder to build agentic workflows that span across products. When we pressed this point at the conference, a senior engineer agreed that the separation adds difficulty to their AI product development.

Given those constraints and how recently we began to see progress in Guidewire’s AI development, we will continue to watch this area closely.

Valuation[1]

We model Guidewire over a decade and assume that global DWP grows in line with long-term nominal GDP. Given the strength of Guidewire’s market position, products, moats and win rates, our base case has penetration of global DWP increasing from 30% to almost 50%. We assume the cloud share of the DWP Guidewire serves rises from 60% or more to almost 90%, and the cloud share of ARR rises from 84% to 95%.

Cloud migrations raise the price Guidewire earns on the same premiums. We assume a further 2% annual growth in pricing in our base case as customers adopt more products. To note, because of the DWP-based pricing, this is not equivalent to a typical fixed contract price increase. The inflation aspect is naturally accounted for already in DWP growth. We then look at the range of pricing and DWP penetration combinations that can deliver our assumed growth, as discussed above.

As growth slows over the decade, we think operating margin can improve in line with that of other strategic enterprise software vendors. We assume a 4% exit free cash flow yield, equivalent to 25 times FCF, reflecting the quality of the business and its growth at exit.

This gives us a 16% IRR from $170, with a downside and upside up 4% and 21% respectively[1].

Business Update

There are no business updates this quarter.

Footnotes

[1] Refer to our disclaimers, we make no guarantees of future performance, and our projections should not be relied upon

[2] Guidewire’s contracts often include fixed price increases. This is especially common in cloud conversion contracts, where early discounts help customers cover transition costs. Fully ramped ARR reflects the same contracts at the agreed higher prices.

[3] We assume a 35% ending FCF margin, 5% terminal growth and a 12% weighted average cost of capital.