Substrate AI — AI venture builder

Substrate AI buys, builds and scales ventures with proprietary AI to transform companies across different sectors. The company identifies cash-generative businesses with the goal of deploying AI to transform their performance and valuation with the technology developed by its core holding, Subgen AI. Subgen AI is an AI software and services business, which works with enterprise and government customers to enable them to deploy AI-enabled solutions and services. Subgen’s key product, Serenity Star, provides clients with a framework for rapidly deploying and managing large language model (LLM) generative AI solutions across their businesses. Serenity Star as a MaaS was launched in February 2024 and is being sold on a subscription basis to organisations. Substrate is planning to separate the Subgen business as a private company and Substrate’s management expects to maintain c 15% ownership of the business, with an exclusive licence to sell Serenity Star in Spain, Africa and the Middle East. Substrate AI’s holdings span healthcare, finance, HR and energy.

There are limited financials for Substrate and its subsidiaries given the early-stage nature of the businesses. The most recent (FY23) results showed a strong, more than doubling of revenue to €8.6m (FY22: €3.1m) as its portfolio companies scale rapidly. Management expects Substrate to generate revenue of €20m in FY24, although this is inclusive of €5m from Subgen AI. To date, Substrate has been loss-making at the EBITDA level, but management expects the group to deliver its first year of profitability in FY24, with c €4m in EBITDA compared with a loss of €3.7m in FY23, although again this is before the transaction. Substrate has stated that, assuming it delivers its maiden year of positive EBITDA, it will not let net debt grow to more than 2.5x EBITDA, with the last reported net debt figure €8.6m (FY22: €1.9m) at 31 December 2023. The last reported net debt figure is inflated due to capital increases payable in shares, which, when adjusted for, gives Substrate an approximate net debt figure of €0.6m. In a recent company presentation, Substrate valued its portfolio on a book value basis at €45.7m. However, this will reduce once the Subgen reorganisation has completed. As the company expects to maintain at least a 15% stake in the private Subgen business, we expect the book value to reduce to €23.4m, assuming Subgen AI is worth the €26.3m book value provided by the company.

The success of Substrate AI is largely dependent on management’s ability to identify companies and successfully execute the integration of AI capabilities within their offering to add material value. These early-stage businesses carry inherent execution risk and, although many of Substrate AI’s portfolio companies are cash generative, any adverse business performance could put a potential strain on Substrate’s liquidity. Furthermore, its focus on AI means it is susceptible to technological obsolescence risk given the extraordinary pace of technological advancement in the industry. Other sensitivities include the lack of public key performance indicators (KPIs) for its portfolio companies, appropriate market conditions for portfolio exits and the lack of exits to date. However, this is to be expected given the early-stage nature of the business.

Based in Madrid, Substrate AI was formed in June 2021, having originally been founded in December 2010 as KAU Finanzas by Lorenzo Serratosa and José Iván García and later known as Zona Value in March 2018. The company is listed on three exchanges, the BME Growth Market in Spain, OTC Markets in the US and the Aquis Exchange in the UK.

We list the acquisitions made to date below:

Exhibit 2 shows the current structure of Substrate AI. As the organogram shows, many of the businesses are already cash generative and therefore do not require additional capital injections from Substrate.

In June 2022, Substrate signed a convertible bond agreement with Alpha Blue Ocean to issue €20m of bonds convertible into shares across three years. To date, 11 tranches have been issued, with nine in FY23 raising €4.5m and three in FY22 raising €1.3m. Furthermore, the company issued 2.7m warrants associated with the bonds in FY23 and 0.6m warrants in FY22, totalling 3.3m at the end of FY23. We detail below the equity raises Substrate has undertaken to date:

Consequently, through the conversion of the bonds, issues through acquisitions and equity raises, management has increased its share count from 24.51m at the end of FY22 to 71.97m Class A shares and 138.64m Class B shares (although these are non-voting) at the end of FY23.

On 25 June 2024, the Substrate AI management team called an extraordinary general meeting for 26 July to distribute the share premium of its largest subsidiary by potential value, Subgen AI. We explore the Subgen AI business in further detail later in this report.

Management has proposed distributing the share premium for a maximum amount of €1.5m in Subgen AI, through which shareholders have the choice to receive shares or an equivalent cash payment. Prior to the announcement, Substrate AI held 80% of Subgen. Management has issued 150m new shares for a nominal value of £0.01 per share, reflecting the share capital of £1.5m. Substrate expects to maintain 30m shares which, after additional distributions, would leave it with a 22m shareholding in Subgen, equivalent to 15%. At the time of writing, Yaro Investment Holding (which is controlled by Fleebe’s CEO Yann Roche) holds 18.1% of Substrate, with JMSAN Agentes Financieros Globales (controlled by Lorenzo Serratosa and Iván García) holding another 17.3%. Consequently, although Substrate is losing direct control of the company, through related parties it will maintain a significant amount of control over the business, hence it will use the equity method of consolidation when accounting for Subgen.

Substrate AI outlined the main reason for the proposed restructuring:

Following the extraordinary meeting on 26 July, shareholders will have from 1 August to 30 September to decide whether to receive shares in Subgen AI or a cash-equivalent payment. Management expects the reorganisation to complete on 7 November. Although there are no guarantees, as some shareholders may take the cash payment option, management anticipates that Substrate AI will maintain a shareholding of at least c 15%. Consequently, Substrate would lose direct control but would maintain a significant stake, although with related parties it would continue to control more than 50% and benefit from any subsequent uplift in valuation following a public listing.

Investors who take up the distribution rather than the cash payment should be aware that the shares would likely be illiquid due to the new Subgen shares being in a private rather than public company.

At time of writing, Subgen AI is Substrate’s largest subsidiary company by both book value and potential valuation against comparable multiples. It provides generative AI services and solutions to clients across a broad range of industries through both its Serenity Star solutions and its LLM system solutions. The Serenity Star product is still relatively new, although it is based on AI patents and technology that the company has developed since 2018. Broadly, Serenity Star seeks to solve complex problems or create efficiencies for businesses and governments by providing a framework for the integration and deployment of generative AI and LLM solutions into their systems. Headquartered in London, Subgen AI has more than 120 employees and serves clients across the EU, the US and Latin America.

The Subgen AI technology has been developed by Bren Worth (brief biography below) since 2018. Bren became part of Substrate AI in 2021, when Substrate acquired his technology for €15m and subsequently invested a further €1.5m to develop the technology to become the base of Serenity Star. Substrate then acquired 70% of software development group Binit, which has more than 100 developers. The majority of these developers are located in South America, enabling Subgen to benefit from a lower cost of labour than developer counterparts in Europe. In March 2024, Substrate acquired a further 21% of the Binit group for €2.1m, satisfied in shares in Subgen AI. Later in March 2024, Subgen signed a contract for an investment round of €10m with GT Securities, a Californian investment bank.

Through both its existing expertise and the software development capabilities offered through Binit, Subgen can provide clients with a broad suite of generative AI services, from data management through to platform usage.

Subgen’s disclosed financials are limited, but in a recent presentation management noted that €5m of revenue was expected by end FY24. As mentioned previously, investors who take up Subgen shares in the offering should note the shares will likely remain illiquid until listing given the shares would be in a private company.

Following the distribution of Subgen’s capital and Substrate reducing its shareholding in the company from 80% to c 15%, the commercial relationship between Substrate and Subgen will change. It is expected that Subgen will provide Substrate with an exclusive licence to sell Serenity Star in Spain, Africa and the Middle East, while Subgen will continue to operate as an independent business. We provide more detail on Subgen’s business below due to its continued importance to Substrate despite the lower level of control, and considering the exclusive licence Substrate has on its products.

Subgen AI is headed by CEO Lorenzo Serratosa and CFO David Jiménez. We profile both Lorenzo and David in the Management section later in this note.

From a technology perspective, the Subgen AI technology team is led by co-CTOs, founder Bren Worth and Mei Si, who together have more than 25 years of experience in the field of AI. The strength of the team has allowed Subgen AI to develop a product that addresses clients’ needs in the fast-evolving generative AI market. We profile the key members of the product development team below:

Subgen’s primary solution, Serenity Star, is a business-to-business (B2B) AI deployment platform. It provides a platform and framework of tools to enable businesses and governments to adopt and manage generative AI and LLM-based solutions.

The platform offering is flexible and scalable, with clients choosing the level of service relative to their requirements and budget. Highlighting the breadth of its offering, the client base spans a range of industries, including manufacturing (John Deere), energy (Canadian Solar), consumer products (Mahou San Miguel), law (Cremades & Calvo-Sotelo), construction (Holcim Group) and government organisations (Buenos Aires Ciudad). Serenity Star represents Subgen’s AI-as-a-service model for clients.

The Subgen AI team has been developing specialised co-pilots, agents, plug-ins and apps within its Serenity Star platform, which act as expert assistants for users. These co-pilots provide data insights, suggestions and strategic advice to users, enabling employees to focus on core operations and driving efficiencies for businesses. A key differentiator relative to peers is the use of multiple, custom or off-the-shelf LLMs, as opposed to a single LLM such as Copilot or ChatGPT, which are powered by OpenAI’s GPT models. This enables Serenity Star’s co-pilots to provide more specific and relevant responses to tasks, built around the company’s existing data sets, while diversifying the risk of over-reliance on a single LLM that may be subject to a system failure or disruption. The service also enables the client to monitor costs by utilising various LLMs, with a cost dashboard integrated into the platform to maximise efficiencies.

Users integrate the Serenity Star platform into their systems, providing company-specific data that a standard LLM most likely would not have access to. The product is low-code/no-code, allowing it to be easily integrated without needing a specialist employee to integrate it. Subsequently, users can create agents focused on a specific task based on Serenity Star’s offering, which spans chat, activity, plan, assistant or co-pilot. Serenity Star has developed proprietary technology to assess the user’s request and determine the most appropriate or efficient LLM to use based on a combination of cost and capacity. The user can also choose a specific LLM they would like to run the prompt through. Once a task has been created, the platform works by connecting to more than 100 LLMs, agents, co-pilots, plug-ins and more to provide the most efficient solution tailored to the prompt. Through this framework, the client can easily monitor the uses of agents created through an integrated dashboard.

The Serenity Star product is cloud based but cloud server agnostic, meaning it can be integrated into a company’s technology ecosystem at the enterprise level relatively seamlessly. Furthermore, the Subgen AI team offers the full cycle of data management for businesses looking to feed LLMs with their data through its software development business, Binit. This helps to streamline the process of data management, which would otherwise be a complex process given varying data forms.

Clearly, Subgen AI’s offering is broad and wide ranging as clients can utilise the platform to solve specific solutions for their businesses. Its platform provides customers with a unique offering through access to a variety of LLMs rather than just one. Exhibit 7 below summarises some key client success stories to showcase some use cases.

Serenity Star’s pricing model is dynamic and scalable. It has two offers: a pay-as-you-go service or a customised plan for businesses. The pay-as-you-go service provides businesses with all the features of the Serenity Star platform, with users getting a free €15 credit trial for three months and with no credit card required. Within this offering, clients pay for a certain number of token credits which correspond to the usage allowance, meaning there is no additional cost to Subgen for additional usage. We believe this pay-as-you-go model will be particularly attractive to early-stage startups, which likely require a low-cost and flexible offering. The customised plan provides clients with an exclusive account manager, priority support and discounts relative to the volume of credits bought. Management believes that its pricing model provides a clear pathway to achieve fast revenue growth in a nascent but rapidly growing market.

Alongside the Serenity Star AI-as-a-service offering, Serenity Star has the capability to build out on-premise and domain-specific LLMs for companies or government organisations. The development of proprietary LLMs is expensive and the costs of development are increasing exponentially. Lambda estimated that to train OpenAI’s GPT-3, which had c 175bn parameters, cost $4.6m, while Forbes estimated that Bloomberg’s BloombergGPT, with c 50bn parameters, cost $2.7m. Subgen estimates that it would cost a company $6–8m to create its own proprietary Serenity Star product. However, Subgen has the capability to develop an on-premise or domain-specific LLM for those companies or governments that wish to build out their own LLMs due to concerns around data sovereignty, privacy, compliance or efficiency. Pricing on these domain-specific LLMs would differ project by project and would depend on the size of the LLM built. Showcasing this, Subgen has already built its own proprietary LLM, Serenity 7B, which the Serenity Star platform also plugs into as one of the many LLMs available on the platform.

Serenity Star’s offering is particularly important in Europe, as new and evolving regulation means there is increasing scrutiny to ensure companies and governments are using AI in a compliant way. The EU AI Act (May 2024) will require companies to assess the risks when implementing AI into their systems. The act uses a risk-based approach related to the applications of AI, categorising risk within four buckets of unacceptable, high, limited and minimal (or no) risk. Penalties for infringements range from the higher of €7.5m or 1% of global turnover to €35m or 7% of global turnover.

Although this provides a framework for companies to think about how they adopt AI, it can make it seem more complicated for those that are yet to integrate AI capabilities. Ultimately, companies utilising AI need to remain in control and evaluate every request being made to avoid misuse of the technology.

Seeking to address the growing issue for companies wanting to utilise generative AI within their systems, in March 2024 Serenity Star launched Serenity compliance, a regulatory and compliance tool based on Subgen’s proprietary LLM. The tool enables companies to monitor whether AI agents are being used in a compliant way relative to either the EU AI Act or, alternatively, a company’s own compliance policy. A company’s specific compliance policy (eg an expensing policy) can easily be uploaded to the platform. The compliance section of the platform provides a dashboard of requests that have violated EU laws or company-specific compliance procedures, enabling clients to interpret and act on the data easily.

Binit is a software development and consulting group operating across Spain, the UK and Latin America with a development centre in Buenos Aires. The group was co-founded by current Binit CEO Juan Pablo Di Pietro (who works for Subgen in business development) and Chief Technology Officer Leandro Harillo (who works for Subgen as chief product officer) and has more than 100 developers. Binit spans several industries including agritech, smart retail, healthcare, community networks, edtech, logistics, shared economies and e-government. The company designs and creates digital products, ecosystems and platforms for clients seeking to develop proprietary technology. It employees the full range of software developers, including project managers, software engineers and quality assurance engineers. Its clients include John Deere, Charles Taylor, Sodexo, DT Global, JBS, Holcim and Buenos Aires City.

Subgen can leverage this software development expertise within its offering, providing clients with the full suite of support. This includes normal IT service functions such as data management or application design, but the company is evolving to serve the AI market. Some of the senior management team, including Leandro Harillo, worked on launching the Serenity Star product.

Subgen initially bought 70% of Binit and Deltanova, the parent companies of the Binit Group, in November 2023 for €6m and acquired a further 21% of the business in March 2024 for €2.1m. The transaction was satisfied in Subgen shares.

We will briefly discuss the growth of the AI market to provide some context to Subgen’s market opportunity. Since the release of OpenAI’s ChatGPT in November 2022, the topic of generative AI has exploded in the public consciousness, while the rate of technological development has been exponential. The potential impact on global GDP is significant, with Goldman Sachs estimating that generative AI could boost global GDP by 7% by 2034 through productivity improvements. The global AI market is expected to grow rapidly at a 17% CAGR between 2023 and 2030 to $739bn (source: Statista). Looking at Subgen’s primary markets of Europe and the UK, the European AI market is expected to grow at a faster 29% CAGR over the same period to $141.8bn (source: Statista). In Spain, Subgen’s local region, Statista forecasts that the AI market will grow at the same CAGR to $12.5bn. To date, adoption of generative AI has been slower in Europe than in the US. However, this presents a significant growth opportunity for Subgen as companies increasingly look to integrate AI solutions into their businesses to drive productivity and growth. Subgen’s offering is particularly attractive in the European context as a result of growing concerns over data sovereignty for those businesses or governments feeding models with sensitive data.

Having examined Subgen AI’s market opportunity and product offerings, we now examine some of the challenges facing the AI industry and how Subgen is seeking to address them. Many of Subgen’s competitors that provide AI platforms at the enterprise level will be much larger in scale, such as C3.ai (market capitalisation: US$3.8bn), and are likely to have greater access to corporates. However, Subgen’s offering, focused on small and medium-sized enterprises (SMEs) and start-ups, should position it well to take advantage of demand from this end of the market.

Subgen AI seeks to sell its services in an AI-as-a-service model, creating recurring revenues as it provides clients with access to hundreds of LLMs. Given the current pace of the AI market and advancements, many companies struggle to understand in the first instance how AI could be leveraged within systems to generate meaningful efficiencies or insights. The Serenity Star platform helps SMEs or governments deploy AI in a manageable way within their own technology ecosystems. The other customer Subgen AI is targeting is start-ups, which seek to utilise AI to scale up their businesses. Many start-ups want to utilise generative AI to accelerate their growth, but this can often be expensive. Serenity Star’s pricing model enables many start-ups to access the platform at a reduced and accessible cost, with the flexible pay-as-you-go offering. Targeting this market, Subgen is partnering with a number of start-up accelerators, through which it can network easily with businesses that are keen to embed AI capabilities into their systems.

Given the rapid speed at which the generative AI market is moving, many of the issues facing the industry relate to supply chain issues. These include issues around data centre capacity and energy usage, with queries from ChatGPT currently c 10x more power intensive than Google searches (source: Goldman Sachs). Goldman Sachs estimates that data centre power demand will grow at a 15% CAGR between 2023 and 2030, while electricity usage from data centres is expected to grow from 3% of US power demand in 2022 to 8% in 2030.

As a SaaS platform and because it is not training the majority of LLMs on the Serenity Star platform, Subgen AI can operate using standard data centres that do not require high-performance computing power or the use of graphics processing units supplied by the likes of Nvidia. Consequently, it avoids many of the current bottleneck issues faced by the industry.

We focus our narrative in this section on Fleebe and 4D Medica, which together account for c 64% of book value (excluding Subgen) and where exits or liquidity events are planned over the next 12–24 months. Over time, we expect the balance of value in the portfolio to evolve as companies develop at different rates.

Fleebe is Substrate’s holding in the HR vertical. The business comprises a growing, profitable international recruitment agency (IFIT Solutions), which is providing the commercial foundation to support the development and commercialisation of an AI platform for talent acquisition and retention. It is Substrate’s second largest holding by book value (€6.175m).

The development of Fleebe’s AI HR platform was initiated by Substrate in 2022, and gained traction with the group acquiring IFIT Solutions in August 2023 for $4.6m. IFIT is a UK-based but international recruitment company, which has given the business scale, domain expertise and access to market. Its founder and CEO, Yann Roche, serves as CEO of the combined entity.

The Fleebe platform is being developed to enable employers to find, hire, discover, promote and empower talent in their organisation. The product was initially conceived to provide a solution to address HR retention and talent management problems, which came to the fore during COVID-19, such as quiet quitting, poor staff retention and mental wellbeing. The application has been built using the Serenity platform and development teams, although management now expects to build its own technology team to support the platform as the business develops.

The Fleebe platform has two key products: Talent Acquisition and Talent Management, providing AI-enabled decision support and workflow platform solutions to corporate (enterprise and SME) and government customers.

Talent Acquisition provides AI to help companies hire top talent, increase diversity and offer a positive candidate experience, facilitating the sourcing, screening and scheduling processes. This product was unveiled at the MWC Barcelona conference in February 2024.

Talent Management enables employers to create a single view of talent across their organisation, facilitating the collection and updating of employee skills using an AI virtual assistant to survey staff. The platform facilitates the storage, analysis and moderation of these records, with AI deployed for tasks such as internal hiring, establishing teams and defining development plans for staff.

The platform is currently being used on a proof-of-concept basis by pilot customers across Spain, with international and domestic companies across the fast-moving consumer goods, technology, banking/financials and pharmaceutical verticals. At the moment the trials are unpaid, with management eyeing formal launch in October/November, at which stage the business expects to migrate customers to paid services.

Management targets customers across enterprise businesses, SME and government. Given the shorter sales cycles and less involved integration requirements, we believe the barriers to entry are likely to be lowest in the SME market and the company is working with the Spanish Chamber of Commerce to support growth in this segment. The company also has an established channel to market in IFIT’s customer base across the UK, Spain, the US, Mexico and Costa Rica.

IFIT, which provides the commercial foundation for Fleebe, is a specialist recruitment consultancy with headquarters in London and operations in Granada, Spain, Costa Rica and Mexico. It provides a range of services, including HR, headhunting and IT contracting, with the latter being the most significant by some margin. It also has the additional benefit of generating recurring revenues. The company has longstanding relationships (eight to 10 years) with a number of large clients and deploys a ‘land and expand’ approach to growth, securing new clients then aiming to grow revenues and secure preferred supplier status through the quality of its services and account management.

The company has developed a strong track record of profitable growth over the past five years, with revenues growing at a 45% CAGR over four years to FY22 and margins consistently in the 13–14% range. In 2023, there was a hiatus in growth due to the cycle of tech layoffs witnessed globally, but management expects to resume a healthy double-digit growth trajectory in 2024.

In terms of its offering and vision, Fleebe’s closest peers are Eightfold AI and OysterHR, both privately held US business. Both these companies are reported by Crunchbase to have been valued at more than $1bn in their last funding rounds, with Eightfold AI raising $220m in June 2021 (with SoftBank as a leading investor), while OysterHR raised $150m in April 2022. Both companies are currently focused on the US and English-speaking countries, although management comments that they are also seeing AI-based start-ups emerging across Spain and the emerging markets in which IFIT operates.

We believe the key milestones/KPIs for investors to gauge Fleebe’s progress are as follows:

Once the customer base has been established, traditional SaaS metrics such as churn, customer lifetime value and acquisition costs will become relevant, although the company is likely at least 18 months away from this stage.

In May 2022, Substrate acquired a 70% stake in 4D Medica, an established value-added reseller of medical imaging equipment, for $1.5m.

Substrate is using this business as the commercial platform to develop and commercialise a generative AI decision support tool for medical imaging, aimed at using AI to help accelerate and improve diagnoses. The company’s initial focus is on the veterinary market, with a view to expanding into the more regulated (human) medical imaging segment in the future.

4D Medica is an independent value-added reseller and systems integrator for medical imaging equipment including radiology, ultrasound, CAT, CT and MRI into the Spanish market. It was founded in 2011 and is based in Valencia, Spain. Until recently, 4D Medica has focused on the veterinary market, being subject to a non-compete clause in the hospital (human) market, which has now expired. The company acquired Diagximag for €3k in May 2023, a distributor of medical imaging equipment (X-ray and ultrasound) in August 2023 to re-establish the business in the hospital segment and build scale. This was followed with the acquisition of a peer, Equipos Hospitalarios for €2.6m also in August 2023. Substrate’s book value for the 64% holding in 4D Medica is €6m.

The company’s focus is supporting the ‘open’ market of independent, smaller vendors as opposed to the more vertically integrated solutions provided by industry majors such as Philips, Siemens, GE and Toshiba. It has been evolving from a hardware reseller towards a more solution-focused business model for some time, as the migration from analogue to digital imagery drives demand for systems integration services, support and maintenance and software.

The company’s AI strategy is to leverage its installed base of c 2,000 customers, to train the system on anonymised imagery and data stored in the picture archiving and communications system (PACS) software to provide ‘Virtual Radiologist’ decision-making support tools. It is developing two levels of intervention:

Management believes there is a strong need for these solutions, to address a significant bottleneck in radiologist capacity through accelerating diagnoses and, in the veterinary sphere, enabling diagnoses by less specialist staff.

Management has indicated that 4D Medica’s AI diagnostics system is being trialled and trained by clients across the company’s 2,000 customers, while Subgen is carrying out product development. Management expects the service to generate revenue in the veterinary market from September, at which point customers using the service will be charged. The approval cycle within the hospital segment is at a more nascent stage and timescales are difficult to gauge – in part because the approval process for medical software is not yet particularly well defined in either Europe or the US. In Europe, the timetable will likely be influenced by what CE class is attributed to the system (see below). Management has indicated that approval for class II and below can be completed relatively quickly – potentially within 12 months – whereas we would expect class III to take substantially longer. It is possible that 4D Medica restricts usage to class II tasks to accelerate the approval and adoption process in hospitals. The CE classes are:

While the development of AI-enhanced medical imaging is a key element of 4D Medica’s growth strategy, it is not the only growth driver for the business. We believe the following growth drivers have the potential to sustain a healthy growth strategy on their own:

We believe that the key milestones/KPIs for investors to gauge 4D Medica’s progress in AI are as follows:

Management has ambitious growth targets for 4D Medica, expecting revenues to grow from an estimated c €4.4m in FY24 to c €6m in FY25 and to more than €16m in FY26. Substrate’s management has stated that the business is currently profitable. If progress continues to plan, Substrate is likely to consider an exit, potentially via listing, within a 12- to 24-month period.

Zona Value is a Spanish language financial website and marketplace for investment products such as funds, pensions, equities, derivatives and exchange-traded funds (ETFs). The partnership between Substrate and Zona Value dates back to 2017 when the two businesses collaborated on the development of a trading system for US ETFs. This trading system has been running for five years.

The Zona Value site features a number of AI agents, developed with the support of Subgen to help clients screen for stocks based on simple criteria such as value or other financial measures, as well as the models developed by Subgen.

Boalvet is an agritech company specialising in applying AI to optimise livestock health and therefore the productivity of dairy farms’ sustainability. Substrate acquired the business in 2018, so it is one of its more longstanding holdings. The platform enables the collection of data from the livestock and other data sources, which is then analysed using AI to predict potential health issues and optimise diet or care. The company was founded to address a growing need among livestock farmers for more sustainable and higher-quality food production. Boalvet operates primarily in Spain, the EU’s second-largest food producer, but has the potential to expand overseas.

Substrate established Save the Planet in April 2023 to encompass projects related to the energy sector and sustainability. The business’s current focus is using AI to support hotel decarbonisation.

At the group level, there are seven members of the senior executive management team, four board members and four members of an advisory council. Substrate’s board consists of Chairman Lorenzo Serratosa, CEO José Iván García and two independent members, Tawhid Chtioui and Christopher Dembik.

Below we provide brief biographies of the key members of Substrate’s executive management team and board, having earlier profiled the key development personnel at Subgen:

Substrate has built a robust portfolio of companies that it is looking to scale through the deployment of AI. The company has embedded AI into the core investment proposition when acquiring businesses, for which the use cases are expected to show significant growth as adoption becomes more widespread. We see the main sensitivities as:

Given its relatively short trading history, Substrate’s financials are limited. Although it has a portfolio of companies, Substrate has so far disclosed a consolidated set of accounts for the group including Subgen. However, given the proposed restructuring of the company, Substrate’s financials will incorporate Subgen’s financials via the equity method of accounting due to its maintained c 15% ownership and continued significant influence over the company.

In FY23, Substrate more than doubled its revenue to €8.6m, up from €3.1m in FY22. This included revenues generated from Subgen. Given the early-stage nature of its businesses, it is unsurprising that the group is loss making at the operating level, recording a loss of €7.5m in FY23 (FY22: €15.4m). Management previously stated that it expected revenues of c €20m in FY24. However, this included the expected €5m in revenue from Subgen. Following the Subgen transaction, we expect revenues to reach c €16m, inclusive of Subgen revenues under the equity method of consolidation, which would reflect 74% year-on-year growth. Management also expects the group to be EBITDA positive for the first time in FY24, anticipating €3m post-Subgen reorganisation compared with a €3.7m loss in FY23.

We show below the breakdown of revenues by segment as reported by the company. FY23 was a period in which Substrate entered into new sectors, including HR (Fleebe), while its existing sectors showed strong growth rates (see Exhibit 14 below). The Matrix segment is the Substrate AI SA holding company, while Technology AI is both the activities in computer consulting and R&D from Subgen AI, as well as R&D activities from Substrate.

Due to the early-stage and loss-making profile of Substrate’s portfolio companies, the group has consistently reported a free cash outflow since inception in 2021. It has grown its net debt position to €8.6m (FY22: €2.4m) as it has grown and required additional funding. When seeking a potential investment, management looks for businesses that are cash generative and, for the most part, profitable, which could accelerate growth with the introduction of AI in their business. Consequently, four of Substrate’s portfolio companies are already profitable and self-funding.

At an investor day in April 2024, management set out its expected funding requirements between 2024 and 2028, by which time it expects to have listed three of its businesses. Management is expecting to require €180m of funding across the four years, of which €30m will be generated by its businesses, €50m through grants and government subsidies, €25m through new debt and €75m through equity financing, which includes proceeds from the expected IPOs. The capital raised through IPOs will go towards those individual businesses, rather than fund future acquisition targets for Substrate. Clearly, the majority of funding will come from additional equity raised, which has the potential to dilute current shareholders. Substrate has stated that, assuming it delivers its maiden year of positive EBITDA, it will not let net debt grow to more than 2.5x EBITDA leverage. Management has noted that Substrate’s current focus is on scaling its existing companies, although its involvement with start-up accelerators and incubators could obviously present some potential early-stage investment opportunities.

We would expect that following the proposed reorganisation of the business, with the holding in Subgen diluting down to c 15%, net assets would reduce by a similar amount to the €1.5m that is being distributed through the share premium.