Tech valuation is enjoying a Goldilocks moment; we fret whether it is “too hot, too cold, or just right”. Companies are concerned about how a deteriorating macro might throttle back AI spending. This pressures valuation. Investors worry about AI ROI and with users still at the AI experimentation stage (Salesforce: “only half Agentforce deals are paid”) with any further economic deterioration. Trophy AI projects are easy to shutter even if they are the front end of a multi-year investment. This pressures valuation. Tech valuation remains a fusion of sentiment and fundamentals – the ‘sentimentals’. While investors may baulk at high valuations, they are less concerned if the share enjoys upward momentum (shares on an upswing mean performance exceeds their benchmark and they get to keep their jobs). The challenge is spotting any sentiment turn. In the UK we have cheap valuations (relative to others and history) and our indicator says that sentiment is turning more positive, March was the low point. The combination of low valuation and positive sentiment suggests better times ahead.

Latest Tech Universe results span Data & Analytics (C3.ai, Salesforce); IoT/Smart (SRT Marine); Gaming (Frontier Developments); IaaS (Beeks); UK Software (Aptitude); IT Infrastructure Services (Datatec, Softcat); AI & Mag 7 (Nvidia); Defence-tech (Cohort). Read on. Data insights (not navel gazing) inform our evolving views on the tech-economy.

Tech drivers and Points to Ponder

1. What’s up with Tech Valuation

The May UK tech sector PE is 17.7x (n = 95) this is shy of the long term average (20.1x). In the US the situation is different; the PE there is 29.7x a premium to the long term, 27.4 x. Other valuation measures tell a similar tale – note UK EBITDA is 14.5x, the long term average is 15.9x. In our view, the valuation makes the UK appeal to domestic but also global investors looking to diversify holdings. But ‘cheap’ alone is not enough to attract long term investors – what is needed is more positive sentiment.

Our tech valuation heatmap shows valuation discrepancies

Pairing Momentum and valuation shows the cohort differences

. . . looking over time suggests that momentum has turned. . .

US and Them: UK remains attractively priced (relative to itself and the US)

Caution: the valuation gap ‘disappears’ when comparing ‘like-for-like’ – UK vs peers PE (x) ratings – (it’s not the venue that matters)

so for better bargains look amongst smaller companies . .

and here is the shopping list: UK tech TSR since 2020

2. The (Pronounced & Speedy) shift in Vector Search market leaders

Research from the Cube and ETR concludes that Vector search has moved from “headline innovation to checkbox feature”.

The shift has splintered adoption across a wide set of vendors. When asked which vector-search engines they are currently using or plan to use, ETR respondents placed Microsoft’s Azure AI Search narrowly in front, followed by MongoDB’s native Atlas Vector Search and an almost dead-heat between Elasticsearch and Redis. A second cluster comprising Google’s Vertex AI, Amazon Kendra, Databricks’ Vector Store, Snowflake Arctic Vector, and Oracle’s OCI AI Search, occupies the mid-teens, followed by IBM watsonx.data (IBM uses Milvus) and specialist standalone offerings such as Pinecone, Milvus, and Cloudflare Workers AI round out a long tail of single-digit intent.

We are reminded of KX and its Kdb vector database. It was the acknowledged market leader in Vector databases (see below) as recently as 2023. However, it too has suffered being gazumped by the new entrants, although its popularity remains just as strong.

The Great Vector database market take-over

3. Agentic AI – Stealing all the headlines

The results file this week was a ‘love-in’ for Agentic AI.

As a quick reminder, agentic AI is about giving AI agency so it can act, learn, and adapt independently. Today, agentic AI is built on rule-based automation with a large amount of ‘human in the loop’ as industries and regulation decide on the machine-to-machine balance when trying to accomplish complex objectives. The roadmap is to solve increasingly complex, multi-step, goal-driven tasks by using advanced reasoning, iterative planning, and adaptability to operate independently or with minimal or no human supervision. As we add in a wide applicability of agenticAI in areas like Customer Service, Enterprise Automation, Healthcare, Military, Self-Driving Vehicles, coupled with demand from the many companies looking to streamline operations, we can (almost) understand why Anthropic CEO Dario Amodei might surmise that AI could wipe out roughly 50% of all entry-level white-collar jobs.

Agentic AI systems follow four steps:

• Perceive: Gather and process data from various sources (e.g., sensors, databases, digital interfaces).

• Reason: Use LLMs or similar models as orchestrators to understand tasks, generate solutions, and coordinate specialised agents.

• Act: Integrate with external systems and tools to execute tasks based on formulated plans, often with built-in safeguards.

• Learn: Continuously refine strategies and actions through feedback, reinforcement learning, and adaptation to new information.

The debate naturally sidles into the realms of ‘reasoning’.

While AI systems started on repetitive tasks they are advancing towards reasoning, despite hallucinations being implicit in the technology. They achieve reasoning by breaking down complex problems into smaller steps (chain-of-thought reasoning), then analysing those chunks of data to make inferences, and draw conclusions. The framework is based on a combination of a knowledge base (structured information about the world) and an inference engine (logic and algorithms to process that knowledge). This is how AI recommends products in e-commerce scenarios, assists with legal reasoning, or controls autonomous robots. However, it relies more on pattern recognition and memorisation rather than true understanding, and lack flexibility and reliability compared to human logic – hence the popularity of the ‘human in the loop’ approach.

Latest Results review

IT Infrastructure Services

Cohort Scenario: This sector has been cautiously emerging from a cyclical downturn following demand volatility in the years following the end of the COVID pandemic. Since H2/24, demand for hardware has picked up across product categories, Day Sales are recovering, but margins (gross and operating) remain under pressure. Positively the cohort is doing more higher margin services revenue, although it will suffer from the utilisation issues common to the IT Professional Services cohort. There are multiple demand-side drivers for this cohort. The most immediate one is Windows 10 end-of-life, and IT security continuing as a top investment priority for organizations as threats of cyberattacks and data breaches continue to grow and from a regulation the EU Network and Information Systems 2 (NIS 2) Directive (a baseline of cybersecurity requirements for the public and private sector) was adopted into local law by EU countries in 2024. However, soundings from the latest batch of analyst conference calls suggests that this effect could drift into 2026. Generally, the cohort KPIs are emerging from a cyclical downturn. As a caution we note that as spend returns, there will be pressure on inventory days which in turn will soak up working capital – and a feature of the latest results is that inventors are increasing.

IT Infrastructure Services: KPI Dashboard

IT Infrastructure Services: Valuation Snapshot

 

Softcat. A Q3 2025 Trading Update informs that Softcat continues to “perform well”. It delivered double-digit Y/Y gross profit and operating profit growth. This was “broad-based across technology areas and customer segments”, supported by the conversion of some larger solutions projects in the period. As a consequence it upgraded FY guidance to low-teens growth in operating profit, from low double-digit previously. Note: The comment on “larger solutions” is interesting because Softcat sells to SMEs (62% of revenue) with a rump of Enterprise clients (23% of revenue) and it has £43.1k Gross profit/customer – any upscaling would increase TAM but also the complexity of the sale.

Datatec. Results were in line with the prior trading update. The new news is that Datatec decided to increase its dividend payout ratio from 3:1 to 2:1, resulting in a 114% increase in total dividends. CEO Jens Montanana commented that this was possible as operationally the increased IT complexity driven by AI and the significant rise in interconnected digital communities is driving infrastructure demand in areas like networking and cyber security. From the divisions:

• Westcon International. Revenue -11.3% Y/Y to US$2.0bn due to changes in revenue mix and a greater percentage of revenue on a net revenue basis. Gross profit increased by 9.4% Y/Y to US$441.2m, with growth “in all regions and a significant increase in Asia-Pacific”. Adj EBITDA +24.7% Y/Y to US$149.9m.

• Logicalis International. Revenue -5.9% Y/Y to US$1.2bn due to a mix change with more net revenue accounted software and services. Gross margin improved due to higher product margins achieved and strong professional services margin. Adj EBITDA +26.9% Y/Y to US$94.0m.

• Logicalis Latin America. Revenue -11.3% Y/Y to US$455.1m with gross profit slipping to US$103.6m from US$117.9m squeezing gross margin to 22.8% from 23.0% last year. Adj EBITDA +59.5% Y/Y to US$20.1m.

• Mason Advisory Limited. Revenue US$37.1m from the three-month consolidated results for FY24: US$9.6m) and EBITDA US$2.8m (three-month results for FY24: US$1.2 million). The division is now consolidated in the Corporate and Management Consulting segment.

There was no material new news on the Strategic Review. The Strategic Review is to address the “persistent gap between Datatec’s valuation and the inherent value of its subsidiaries”. Datatec broadened its investor relations programme during FY25 and was admitted to the OTCQX trading platform in the US to increase international investor access. There was also a share repurchase programme, started 28 November 2024.

Note: There was an accounting policy change (IFRS 15) for recognition of Westcon International’s software revenue which changed to a net basis (i.e. moving rev/rec to Agent (net) vs principle (gross)). There was no change to accounting policy for Logicalis International and Logicalis Latin America. For current trading there was a message of the continuing trend toward higher software sales and annuity services which improves Group margins and cash flow.

IoT/Smart

Cohort scenario. The Internet of Things (IoT) industry is substantial and growing rapidly across a number of industries, as use cases expand. The IoT market is projected to grow from US$945.6bn in 2024 to US$1,377.8bn by 2029 as the installed base of IoT connections reaches 20bn this year and 31.5bn by 2030. Central to the development are technologies such as Wi-Fi, Bluetooth, 5G, and Low Power Wide Area (LPWA) networks, and end markets in healthcare, automotive, smart homes, security, and industrial applications. However, the (investor) attention caravan has moved to GenAI and so ignores the many opportunities in this cohort. The companies are at the development stage (so R&D/Revenue is a monitored KPI) as we try to understand the Life Cycle positioning (a typical S-curve for a product based business). While this sacrifices profitability to establish product and customer moats, pleasingly, FCF margins have recovered from the H1/24 decline (see below). The challenge now is to recognise the eventual leaders from the mere market participants. There is lots of shareholder value, and value traps, in the cohort.

IoT/Smart: KPI Dashboard

IoT/Smart: Valuation overview

 

SRT Marine. SRT Marine Systems (maritime domain awareness systems) announces that the Indonesian inter-government project finance loan agreement and other contract condition precedents have been finalised. This signals the start of the €167m National Maritime Security System, NMSS, project with the Indonesian Coast Guard (Bakamla). The first shipments commenced on the news. The project is to be implemented in multiple phases over a period of two years followed by eight years of ongoing support, maintenance and data services – giving long term confidence in forecasts and strengthening the investment case.

The NMSS project is the delivery of an initial SRT-MDA System, to establish an advanced integrated national maritime surveillance, command, control and communication capability for the customer. It will enable continuous and persistent real time and near real time surveillance of the entire EEZ and all territorial waters. In addition it provides SRT proprietary AI driven intelligence and insight on maritime activities such as illegal fishing, smuggling and maritime safety. The SRT system also provides a national marine data integration platform which will interconnect all national marine agencies.

The project sees multiple command centres, strategic multi-sensor coastal surveillance towers, long range drones and patrol vessel surveillance integration, all of which are fully integrated as an SRT-MDA system as a single maritime C5iSR system all being established.

Simon Tucker, CEO, commented that the “project delivers a sovereign national capability owned and operated by the government of Indonesia that enables them to build up a massive national dataset within a system they own, control and operate that has highly configurable functionalities to provide deep insight. Furthermore, the flexible AI powered architecture of the SRT-MDA System enables them to continue to grow and evolve the system in the future alongside their ever-changing operational needs and scale.” While these super large contracts can at times be a bane is not managed properly, getting referencability on the project alone should hasten other pipeline deals.

The numbers we track: SRT Marine

IaaS

Cohort scenario: We are in the midst of a large ‘cloud’ shift which favours the IaaS cohort. While the hyperscalers seem the obvious ‘winners,’ the IaaS segment includes an array of companies who will gain from vertical market focus, but also in the wider cloud migration, as well as the more ‘classic’ RAS (reliable, available, Secure) corporate requirements. In addition, most modern applications were designed to run on cloud, rather than on-prem systems. While the newsflow tells us that CapEx is on the increase, there is less attention to improving Adj EBITDA margins and recovering FCF margins which feature in the KPI dashboard.

IaaS: KPI Dashboard

 

Beeks. Beeks’ latest Exchange Cloud deal is with the Australian Securities Exchange (ASX) and forms the basis of the new ASX ‘Colo OnDemand’ service. By removing the burden and capital costs of managing IT infrastructure, Colo OnDemand expands TAM by enabling underserved buy-side firms, niche and smaller brokers, investment firms and emerging software providers to access Australia’s financial markets. This is the second revenue-share contract for Exchange Cloud. While we leave forecasts unchanged, these deals underpin our FY26 estimates. We maintain our view: Beeks is gaining from industry tailwinds (cloud adoption, compliance, data sovereignty, cybersecurity, sustainability, analytics, payment modernisation, AI in risk management), which have enabled it to carve out a moat as market leader in cloud infrastructure for financial markets and payments. In volatile markets, we believe investors are best advised to lean into those companies with a track record of cash generation. We remind that Beeks is a long-term secular growth investment play, and its ‘best of breed’ financial model provides resilience in these uncertain times.

Read more from our coverage at Progressive Equity Research.

The numbers we track: Beeks

Source: Company data, Technology Investment Services

 

Data & Analytics

Data & Analytics: KPI Dashboard

Data & Analytics: Valuation snapshot

 

C3.AI. C3.AI delivered a breakthrough ‘beat and raise’ print. Enthusiasm was slightly marred as the analyst call was too much commentary about sell-side and short funds getting it wrong (enough already), and this detracted from the developing investment case. Q4 revenue +26% Y/Y to US$108.7m, the gross profit margin is 62% with the non-GAAP gross profit margin 69%. In FY25, C3 AI generated revenue across 19 different industries, with non–Oil & Gas revenue accelerating by 48% Y/Y. It also closed 193 agreements through its partner network +68% Y/Y and this accounted for 73% of Group. Illustrating increased momentum in Q4 partner supported bookings grew by 419% Y/Y. The joint 12-month qualified opportunity pipeline with partners increased by 37% Y/Y.

We noted the strong Federal market sales momentum: In FY25, C3 AI closed 51 agreements across the Federal sector, accounting for 20% of total bookings. In Q4, it entered into new and expansion agreements with the U.S. Department of Defense, the U.S. Intelligence Community, the U.S. Air Force, Marine Corps, Navy, the Defense Counterintelligence Security Agency, the Missile Defense Agency, CAE USA, the Royal Air Force, Austal, and Thales, among others.

The C3 Generative AI business revenue grew >100% in FY25 as C3 AI closed 66 initial production deployment agreements for C3 Generative AI, across 16 different industries with 14 C3 Generative AI initial production deployment agreements in Q4. Thomas M. Siebel, Chairman and CEO, C3 AI. “The Enterprise AI market is converging toward AI applications — a category we created and continue to lead. We continue to innovate, most recently in agentic AI. We continue to grow, expanding our partner ecosystem and global reach. And we are ready to scale as demand for AI applications accelerates.”

Salesforce. Here too a very strong quarter and another earnings raise. In addition the Informatica acquisition increases TAM, as it better positions Salesforce in the evolving wider enterprise software market. We like the deal. Here too, agenticAI is making the running. First quarter revenue US$9.8bn, +8% Y/Y with Subscription & Support revenue US$9.3bn +8% Y/Y. Current RPO, US$29.6bn, +12% Y/Y and accelerating growth (accelerating growth is the bug bear with the company.) FCF, US$6.3bn, +4% Y/Y as it returned US$3.1bn to shareholders – the company says the last couple of years were about margin and FCF growth, now hopes to do more growth (following an enhanced product insertion rate). Of the business highlights, we noted:

• Salesforce has closed over 8,000 deals since launching Agentforce, of which half are paid. On help.salesforce.com, Agentforce has handled over 750,000 requests, cutting case volume by 7% Y/Y.

• Comments of strong revenue growth in the UK.

• In addition to ‘ADAM’ through the conference call, there was much commentary on cross-selling though the product suite.

Marc Benioff, Chair and CEO: “We’ve built a deeply unified enterprise AI platform—with agents, data, apps, and a metadata platform—that is unmatched in the industry. With Agentforce, Data Cloud, our Customer 360 apps, Tableau, and Slack all built on one trusted, unified foundation, companies of every size can build a digital labor force—boosting productivity, reducing costs, and accelerating growth.”

The numbers we track: C3.ai, Salesforce

UK Software

Cohort scenario: These companies are typically smaller than their US counterparts, but all sell and compete globally. In addition to being smaller they typically have poorer KPIs (notably cash-based, ARR, NRR) but have better profitability. The comparative KPI mix may on its own account for the lower valuation and support the view that ‘venue’ has little to do with the valuation. The group is cheaper and pays a dividend/income in addition to share buybacks.

UK Software: KPI Dashboard

 

Aptitude Software. AGM Statement and Board Appointment featuring:

• 2025 performance will be in line with expectations, despite continued economic uncertainty, particularly in North America.

• Three new enterprise wins year to date for Fynapse, Aptitude’s lead AI autonomous finance solution.

• Organisational transformation to partner-led model now substantially complete, providing the foundation to unlock the full value of the Fynapse opportunity.

• Continued cross-sell and up-sell successes, securing a number of renewals with existing clients.

• Appointment of Paula Dowdy as Non-Executive and Senior Independent Director. NED Barbara Moorehouse departs.

While the macro-economic environment continues to have some impact on corporate decision making, particularly in North America, increasing partner engagement is driving an improved Fynapse pipeline. This is supported by continued renewals, cross-sell and up-sell successes with existing clients in the period, leading to an expanded footprint within the installed base, reinforcing the long-term opportunity for future Fynapse adoption.

Following this, Aptitude announced the renewal of its £20m share buyback (initially announced on 21 March 2024) of which £6m has been repurchased to date. The Share Buyback Programme will be for on market purchases of up to a further £6m.

The numbers we track: Aptitude Software

Gaming

UK Gaming: KPI Dashboard

Gaming cohort: Valuation snapshot

 

Frontier Developments. Hot on the heels of Trufin, news from Frontier Developments is that Trading (end 31 May) sees revenue to be c. £90m (FY24: £89.3 million), slightly ahead of expectations (£85m – £89m) due to “strong performances across Frontier’s game portfolio”. Revenue was boosted by a £3.5m gain on the sale of the publishing rights for the Foundry game Stranded: Alien Dawn. This is not accounted for as revenue but ensures better-than-expected profitability and a higher-than-expected cash position. More news in a Trading Update, 11 June.

We are (sadly) not out of the woods yet. This week EA announced that it cancelled its Black Panther game and closed Cliffhanger Games. According to EA the moves are to sharpen “our focus and put[s] our creative energy behind the most significant growth opportunities.” Less than a month ago, EA cut 300 jobs. The redundancies chiefly impacted developers at Apex Legend developer Respawn and EA’s Experiences team. The game ‘Clair Obscur: Expedition 33’ send some valid messages to the wider industry. It sold 2m copies within 12 days of launch it is an original title made by a relatively small team. For us, this gives hope to the wider games industry as it ponders its next steps in the face of a saturated market, widespread layoffs and spiraling development costs.

The numbers we track: Frontier Developments

AI, Hardware & Semi & Mag 7

Nvidia – Many paths lead to its Valuation snapshot

 

Nvidia. Q1 was impacted by the US export controls to China and while this cost Nvidia c.US$8bn in lost revenue, the forecast guidance still met consensus. Q1 revenue US$44.1bn +69% Y/Y. The 9 April export controls impacted sales of the H20 and NVIDIA incurred a US$4.5bn charge associated with H20 excess inventory. It is also unable to ship an additional US$2.5bn of H20 in the quarter. The GAAP gross margin was 60.5% but excluding that irksome US$4.5bn it was 71.3%. Jensen Huang, founder and CEO: “Global demand for NVIDIA’s AI infrastructure is incredibly strong. AI inference token generation has surged tenfold in just one year, and as AI agents become mainstream, the demand for AI computing will accelerate.” Also Data Center revenue, US$39.1bn, +73% Y/Y illustrate of continued AI demand. Responding to tariffs, Nvidia announced that it is building factories in the US and working with its partners to produce NVIDIA AI supercomputers there.

Nvidia shares have suffered from concerns about trade tensions, how tariffs might impact data center expansion and so create headwinds for AI chip demand. However the concerns dissipated with the print and the very favourable market commentary despite TAM reduction due to closing off the China market.

Other points of note:

• First-quarter Gaming revenue was a record US$3.8bn, +48% sequentially and +42% Y/Y. This adds to our ‘gaming is coming back’ viewpoin

• Nvidia disclosed that it is facing scrutiny inside China, where regulators have demanded that it keep supplying local companies in return for regulatory approval of its acquisition of Mellanox. “If regulators conclude that we have failed to fulfill such commitments or we have violated any applicable law in China, we could be subject to financial penalties, restrictions on our ability to conduct our business.”

• We had surmised that the company would be hit by increased inventory and so impact working capital. This was an error. Both continued to improve.

Some of the Nvidia numbers (US$bn)

Nvidia: Slide of Awesomeness

Defence-Tech

Defence-Tech cohort: Valuation snapshot

 

Cohort. Continuing the drum-beat of positive noise from the Defence-tech sector, an FY Trading Update from Cohort says that the year (ended 30 April) concluded in line with market expectations (Revenue: £245m, Profit: £27.6m, Adj EPS: 46.1p). Among the highlights:

• Net funds, £5+m, “significantly ahead of expectations”.

• Order intake, c.£285m (excluding c.£80m order book acquired with EM Solutions) compared with £387m a year ago. As 2024 included an exceptionally large Royal Navy order, £135m, the like for like is +12% Y/Y.

• The closing order book, c.£615m, marks a new record for the Group, £518.7m last year.

The result was driven by excellent organic growth in the Communications and Intelligence division, supplemented by an initial three-month contribution from EM Solutions, with the performance of the Sensors and Effectors division broadly flat Y/Y. The Group net margin was 10.2% (2024: 10.4%).

Outlook for FY26. Cohort continues to see good demand for products and services from both domestic and export customers. The drivers for investment in defence remain strong, with the ongoing conflict in Eastern Europe and continuing tensions in the Indo-Pacific region leading to increased global defence spending. Cohort awaits the findings from the UK’s Strategic Defence Review, but expects this to maintain a focus on technologies and capabilities aligned with what the Group provides.