Summary
- MH is an active Belgian micro-company founded in 2007, with one registered establishment and one current director; its public accounts summary reports no full-time workforce figure, making key-person concentration more important than corporate breadth.
- The company holds Local Internet Registry status, an IPv4 /22 containing 1,024 addresses and an IPv6 /32. The IPv4 block is operational and RPKI-valid, but it is originated on the public Internet by Datacenter United's AS42160 rather than by an autonomous system controlled by MH.
- The IPv6 allocation was not visible in global routing data on 10 July 2026, while MH's website and three listed name servers all resolved inside the IPv4 /22. That is evidence of useful address control, but not of an independent, dual-stack network with diverse failure domains.
- Public financial summaries show 2024 gross margin of about negative EUR 1,492, EBITDA of about negative EUR 2,234, a net loss of about EUR 20,011, debt of about EUR 99,221 and negative equity of about EUR 13,910. Revenue was not disclosed, so growth cannot be separated from value creation.
- MH's strongest commercial case is not generic cloud hosting. It is the combination of custom business software, hospitality workflows and embedded electronics, especially where a customer values local adaptation, stable addressing and integration with existing equipment.
- Scaled substitutes are severe: Odoo bundles broad business applications and cloud operations at per-user prices, while Lightspeed advertises Belgian restaurant plans from EUR 89 to EUR 249 a month and serves roughly 144,000 locations globally. MH publishes neither prices nor customer references that would demonstrate a superior acquisition or retention model.
- The explicit judgment is that MH remains an infrastructure price-taker until durable contracts, customer diversification, independent continuity controls and recurring gross margin prove that its resource-holder footprint produces economic rent rather than fixed cost.
Relevance is the incentive; scale is the constraint
The starting point is not whether MH possesses technology. It plainly does. The question is whether the company can turn that technology into cash flows that survive the bargaining power of suppliers and customers. A small operator can own scarce addresses, maintain software and understand hardware while still creating little economic value if every sale requires fresh engineering and every month begins with fixed obligations to a registry, a data-centre provider, software vendors, accountants and security services.
Management has three plausible motives for keeping the resource footprint. First, stable address space can support hosting, name services, allow-listing and customer continuity without renumbering every time a retail access provider changes. Second, it can reinforce a local-control proposition for Belgian customers that dislike placing every workload behind a global cloud account. Third, IPv4 scarcity gives the block some transfer optionality. None of those motives is irrational. The mistake would be to confuse an option with an operating advantage.
The economic test is therefore demanding. Who pays MH for control that cannot be obtained more cheaply from a standard cloud or software subscription? How much of that payment repeats without a new project? What service commitment does MH accept in exchange? Which supplier receives a fixed payment regardless of MH's utilisation? And when a founder-led implementation needs emergency support, who carries the downside: MH, its upstream, or the customer whose restaurant, factory or administrative process has stopped?
Public evidence gives partial answers. Customers appear to pay for tailored software, integrations and electronics. MH benefits when modules can be reused across engagements. Datacenter United and other service providers benefit from recurring infrastructure demand. The owner appears to carry financing and key-person risk, while customers may carry switching and continuity risk. What is missing is evidence that recurring contract value is large enough to make that allocation of risk sustainable.
The legal company is smaller than the product surface
The Belgian Crossroads Bank for Enterprises identifies MH as enterprise 0890.151.479, active since 18 June 2007 and in normal legal standing. Its registered office is in Jabbeke, its legal form is a private limited company, and it has one establishment unit. The current register names Matthias Barremaecker as director and also preserves the historic designation of business head, which Belgian law now reads as a directorship.
A 2023 corporate filing sharpens the control picture. It records a single shareholder, the conversion of previously unavailable equity into distributable equity, a move of the registered office, and the reappointment of Barremaecker as a non-statutory director. The same filing gives the company an exceptionally broad corporate purpose: domain names, email, web and server hosting, databases, software, security, consultancy, training, computers, electronic equipment, printed circuit boards, wood, plastics, property and financial investments. That breadth grants legal freedom. It does not show that each line produces sales.
The operating evidence is narrower. The enterprise register lists computer infrastructure, data processing, hosting, computer consultancy and facility-management activities. The company website presents custom software, a hospitality product called RestoCore and custom electronics. Every public contact path points to the same Jabbeke address and telephone number, and prospective customers are directed to Matthias personally. There is no public team roster, second office or named service desk.
The public account summaries classify MH as a micro-company and report zero full-time equivalents or no available workforce figure. That wording matters: it should not be read as proof that no one works for the business. It does mean the public evidence does not support an employed delivery organisation with visible depth. A founder can use contractors, suppliers and automation. But those choices substitute variable external cost for headcount; they do not remove capacity limits, documentation needs or succession risk.
The actual boundary is thus a single Belgian company with long tenure, one visible decision-maker, proprietary technology claims and outsourced infrastructure. That is a viable shape for a specialist studio. It is not the shape of a regional carrier, a scaled software vendor or a data-centre operator. Strategy should begin by accepting that boundary.
MH sells adaptation, not generic software
MH's website makes a coherent commercial claim: it starts from reusable modules and adapts them to the customer's workflow, avoiding both a blank-sheet build and the compromises of an off-the-shelf package. Its in-house iCore framework supplies multilingual interfaces, role-based permissions, data-level access controls, API keys, tags, reporting and geofencing. A separate development environment is said to maintain modular versions and preserve data through upgrades.
That proposition can earn value. Custom software buyers rarely pay for code alone. They pay to avoid changing a process that embodies local knowledge, to connect old equipment, to reflect unusual pricing rules, or to keep a workflow that a broad platform handles badly. Reusable foundations can make this profitable when the price reflects the avoided disruption and when each new customer contributes improvements that can be reused again.
MH's published modules cover customer and supplier records, actions and reminders, contracts, purchasing, product structures, multilingual data, quotations, delivery, invoicing, bank-file reconciliation, reporting and online payments. The invoicing offer includes Peppol support. The company says it does not charge licence fees per user. This is a direct answer to the per-seat economics of larger business suites.
But a zero per-user fee is not itself a business model. MH still has to recover analysis, implementation, migration, hosting, support, upgrades, security and custom integration costs. If the customer has five users, a per-seat discount may be commercially irrelevant. If it has 200 users, the foregone licence charge can become a meaningful advantage, but only if MH can support a 200-user environment and price the project accordingly. Without public pricing, contract terms or deployment sizes, the claimed advantage cannot be measured.
The strongest reading is that MH sells a hybrid of project work and reusable intellectual property. That can produce high gross margin when scope is controlled and support is standardised. It can also become disguised labour, with every client variation creating a permanent maintenance obligation. The accounts suggest the second risk has not been conquered.
RestoCore narrows the problem but inherits hospitality risk
RestoCore is the clearest attempt to convert broad custom capability into a vertical product. It centralises recipes, product structures, stock, suppliers, purchasing and cost calculation for restaurants, bistros and caterers. It can represent a wine bottle and its glasses, allocate ingredients to recipes, calculate waste factors, maintain channel-specific prices and update stock from sales. This is credible domain specificity rather than a generic list of software features.
The product also takes a pragmatic position on the point of sale. MH does not claim to build the till. RestoCore connects to the system a restaurant already uses, provided that supplier exposes integration capability. Products, prices and availability move to the till; sales return for stock adjustment and analysis. That reduces the disruption of adoption and lets MH compete above the payment terminal rather than against every terminal vendor.
The same choice creates supplier dependence. A till provider can change its interface, rate limits, commercial terms or certification rules. A restaurant may replace its point of sale and expect MH to preserve continuity. A supplier may not expose the data needed for two-way synchronisation. Each connector can therefore look like a reusable asset while behaving like a separate maintenance liability.
Hospitality also shifts the credit risk toward the software supplier. Belgian official statistics record 166 accommodation and food-service bankruptcies in January 2026, 169 in February, 206 in March, 186 in April and 125 in May. That is 852 in five months. The sector has genuine need for recipe costing, purchasing discipline and margin visibility, but need is not the same as ability to pay. A weak restaurant can value RestoCore highly and still cancel, delay implementation or default.
The commercial response should be visible in contract design: an implementation payment that covers onboarding, a recurring fee that covers hosting and support, paid connector changes, clear data export, and no implicit financing of distressed customers. None of that is disclosed publicly. Until it is reflected in recurring gross margin, the vertical focus may concentrate MH in exactly the customers most likely to produce support intensity and payment volatility.
Electronics creates differentiation and a second obligation
The electronics offer is potentially more defensible. MH describes modular hardware and software built around a microcontroller system, with communications over RS-485, RS-232 and TCP/IP. It says later hardware structures can be added to an installed network and that a generic control layer handles settings, actions, input validation and version changes. Those are concrete engineering claims and fit the broad corporate authority to build and repair electronic systems and circuit boards.
Software plus electronics can solve problems that standard enterprise applications do not. A customer may need a sensor, control panel, serial device, local network and administrative application to operate as one system. The value lies in making the physical and digital pieces reliable together. Global SaaS vendors usually do not want that edge case; large industrial vendors may price it for much bigger deployments. A small local engineer can occupy the gap.
The cost burden is also different. Hardware introduces component procurement, inventory, prototypes, test equipment, field installation, repair, firmware support and product-lifecycle commitments. A component substitution can require engineering before it generates a cent of new revenue. If custom electronics are sold in low volumes, purchasing power is weak and certification or documentation expense is spread over few units. Customers also expect a replacement path years after delivery.
This is why revenue growth would be an inadequate measure. A hardware project can increase invoices while consuming cash for components and outside work. A bespoke software project can book revenue while adding years of support. Value is created only when price covers the full life of the obligation and reusable modules reduce the next project's cost. MH's broad capability becomes attractive when software and electronics reinforce the same narrow use case. It becomes expensive when the company accepts unrelated projects merely to keep utilisation high.
The address estate is real but network control is rented
The network evidence is specific. RIPE NCC lists "MH" BV as a Local Internet Registry in Belgium, linked to the same enterprise number, address and contact identity found in the Belgian register. The organisation record was created in February 2013. RIPE allocated 185.19.180.0/22, a block of 1,024 IPv4 addresses, and 2a00:48e0::/32, an IPv6 allocation large enough for extensive internal assignment.
MH has not left the IPv4 block dormant. On the access date, the company's website resolved to 185.19.181.21. Its authoritative name servers were ns10.mh.be, ns11.mh.be and ns12.mh.be, resolving to 185.19.180.10, 185.19.180.11 and 185.19.181.12. RIPE records also show reverse zones maintained for the four component /24s. That is evidence of operational use and administrative control.
The public route tells a different story about autonomy. RIPEstat showed the aggregate /22 originated by AS42160, held by DC STAR and operated in the Datacenter United environment. The route object is described as MHDCO and maintained by Datacenter United's technical maintainer. It was first observed in April 2013 and was visible to 326 of 327 relevant RIPE RIS peers on 10 July 2026. Its Route Origin Authorisation was valid for AS42160.
This is a sound arrangement for a small customer of a data-centre network. The prefix is globally visible, route-origin security is in place and an experienced upstream handles external reachability. It is not independent routing. No autonomous-system record tied to MH's RIPE organisation was found, and no more-specific route divided the block between multiple origins. MH controls the registration and use of the addresses; Datacenter United controls the visible origin and therefore a critical part of reachability.
That distinction determines who captures value. Customers may value stable addresses and a Belgian operator. MH may avoid renumbering and can place services within its own allocation. Datacenter United can charge for colocation, transit, security and support because the address block does not reach the Internet by itself. The scarce resource improves MH's bargaining position at the margin, but it does not eliminate the upstream.
One routed /22 does not make an independent network
It is tempting to treat resource-holder status as a proxy for an Internet service business. The evidence does not support that leap. The current company website does not advertise retail Internet access, IP transit, autonomous routing, wavelength services, colocation or a network service-level agreement. The Belgian telecom regulator's May 2026 list did not contain a match for MH under its legal name, enterprise number, address or visible director. Belgian law requires notification before carrying out most operator activity on the market.
Absence from that list is not an accusation. It is evidence that the safer classification is a technology company using number resources, not a publicly notified telecom operator. A Local Internet Registry can support its own services or customers without becoming a mass-market ISP. The /22 is an input to MH's operations; it is not proof of carrier revenue.
The physical diversity is also unknown. Three name-server labels can give the appearance of redundancy, but all three resolved inside the same /22 and therefore shared the same visible origin. Public records do not establish whether the servers occupy separate rooms, power feeds or sites. Datacenter United advertises carrier-neutral facilities, redundant paths, 24-hour support and a national Belgian footprint, but those are the upstream's capabilities. MH does not publish which of them it buys.
For customers, the relevant questions are contractual. Are backups outside the origin network? Can DNS survive a failure affecting the /22? Is there a second site? Who answers incidents overnight? Can the addresses move to a new origin quickly? Does MH control the server configuration and encryption keys? Resource registration answers none of these. A credible continuity offer would turn each into a documented commitment.
Without that evidence, the network footprint should be valued as control over naming and addressing, plus potential portability, rather than as a vertically integrated service platform. The company is below cloud scale and below carrier scale. Its economics depend on using the resource to make higher-value software and integration contracts stickier, not on pretending the resource itself is a network business.
The silent IPv6 block is a modernisation warning
The IPv6 allocation adds another contradiction. MH holds a /32, yet RIPEstat found neither an origin nor a visible more-specific route for it on the access date. The main website did not return an IPv6 address. The public-facing services examined therefore relied on the IPv4 estate even though MH has held IPv6 space since 2013.
This does not prove that IPv6 is unused on private systems. It does show that the public evidence fails to demonstrate dual-stack delivery. For a small application supplier, that may not block sales today. Many SME customers remain operationally IPv4-heavy. But the gap weakens the claim that number-resource ownership reflects modern network capability rather than historical optionality.
The economic issue is not fashion. IPv6 reduces dependence on scarce IPv4 for new endpoints, while customer-owned IPv4 remains useful for compatibility and allow-listing. A supplier that can operate both has more deployment choices. A supplier that leaves IPv6 unrouted pays membership and administrative attention without converting the allocation into service utility.
There may be rational reasons: upstream configuration cost, limited customer demand, legacy applications, security work or a deliberate decision not to expose services. Those reasons should be tested against a simple capital-allocation rule. If enabling and supporting IPv6 costs more than customers will pay or more than it saves, delay is sensible. If the company markets infrastructure control, however, a twelve-year-old invisible allocation becomes a credibility gap.
Scarcity creates optionality, not recurring cash
IPv4 has economic value because the free supply is exhausted and cloud platforms now charge explicitly for public addresses. Amazon Web Services lists a price of USD 0.005 per public IPv4 address per hour, equivalent to USD 43.80 per address for a full year, and exempts customer space brought into its environment. That does not mean MH saves 1,024 times that amount: only deployed addresses would otherwise incur the charge, and moving the entire block to a hyperscale cloud may not be practical or desirable. It does show that address control has a measurable rental substitute.
Transfer-market data offer a second reference. One broker's late-2025 report placed average transaction prices across block sizes in the low USD 30s per address, while its 2026 reports described firmer demand and rising prices in several sizes. Applying a low-30s indicator to 1,024 addresses gives a headline amount in the low USD 30,000s. That is not a valuation. It ignores broker spread, block reputation, taxes, legal terms, transfer policy, customer disruption and the loss of operating utility.
RIPE policy does permit eligible resource transfers and records completed transactions. A sale would still be a one-time release of an asset-like right, not operating profit. It might improve liquidity while weakening hosting continuity and the company's local-control proposition. Leasing space could produce recurring income but would add abuse, reputation, support and compliance risks. At a founder-led company, those risks can consume more attention than the lease earns.
The rational use of the /22 is therefore strategic, not speculative. Keep enough space to support valuable contracts, demonstrate clean use and maintain portability. Consider any monetisation only against the full cost and risk. Do not use the block as an excuse to subsidise low-margin hosting. The address estate can support enterprise value when customers pay for the continuity it enables. It cannot repair negative gross margin on its own.
The accounts show activity without durable value creation
MH's most recent publicly summarised financial year is 2024. Because the micro-company filing does not disclose turnover, the central question cannot be answered with a revenue growth rate. That absence is itself important. A company may invoice more while buying more external services and consuming more owner time. The available measures have to be read as evidence of value creation rather than scale.
The 2024 summary reports gross margin, in the Belgian filing sense of value added after external purchases, at about negative EUR 1,492. It had been positive by about EUR 14,630 in 2023, after roughly negative EUR 37,945 in 2022 and negative EUR 38,918 in 2021. Three negative years out of four do not describe a repeatable engine. The single positive year did not establish a durable base.
The same summary reports 2024 EBITDA of about negative EUR 2,234 and a net loss of about EUR 20,011. Cash was about EUR 10,850, operating cash flow about negative EUR 2,715 and capital spending about EUR 4,260. Total assets were about EUR 85,311 against debt of about EUR 99,221, leaving equity at about negative EUR 13,910. Debt rose from about EUR 86,000 in 2023 while equity fell from positive EUR 6,101.
These are small absolute amounts, which cuts both ways. A few good contracts could change the picture quickly. A single delayed payment, failed hardware project or support incident could also consume a meaningful share of cash. Negative equity is not proof of imminent failure, and the legal register continues to show an active company in normal standing. It is proof that the buffer belongs to creditors and future earnings, not to accumulated shareholder value.
The separation between revenue and value creation is stark. Without revenue disclosure, no one can say whether MH shrank, grew or merely changed mix. What can be said is that 2024 external purchases and operating economics did not produce positive value added, financing and depreciation then deepened the loss, and the balance sheet ended with liabilities above assets. Any strategy that adds fixed infrastructure cost before securing contract demand would be poorly timed.
No visible price architecture closes the arithmetic
MH publishes capabilities but no prices, implementation ranges, support tiers, response times or minimum contract terms. Custom work often requires a quote, so the absence of a price card is not unusual. The missing architecture still makes it impossible to test whether management has priced the full obligation.
A sustainable contract would separate at least four economic components. Discovery and implementation should pay for initial analysis, configuration and migration. Recurring service fees should pay for hosting, monitoring, backups, security, registry overhead and routine support. Change requests should pay for customer-specific development and connector maintenance. A continuity premium should pay for response commitments, redundancy and the capacity to intervene when nothing new is being built.
The no-per-user claim can sit inside that structure, but it cannot replace it. Belgian labour is expensive: Eurostat estimated average economy-wide employer labour cost at EUR 48.20 an hour in 2024, among the highest in the European Union. Specialist engineering can differ materially from that average, but the benchmark shows why a local supplier cannot win by selling hours cheaply. Even owner labour has an opportunity cost if it is not recorded as payroll.
The annual RIPE contribution is EUR 1,800 per Local Internet Registry account in 2026. Against a large operator's budget, this is trivial. Against MH's 2024 cash position it is roughly one-sixth before connectivity, servers, backups, software, insurance or accounting. The registry fee is not the main problem; it illustrates how many apparently modest fixed costs can accumulate below scale.
The required commercial proof is modest and concrete: annual recurring revenue, gross retention, gross margin by service line, support hours per customer, implementation payback and cash collected before project expenditure. None is public. The accounts say the current mix has not covered the economic burden consistently. Until price architecture appears in results, the company's strategy remains a set of capabilities rather than a demonstrated allocation of capital.
Suppliers are paid before MH proves demand
MH's supplier stack has several layers. Datacenter United sits at the network edge because AS42160 originates the /22. Colocation or server providers supply power, cooling, physical security and intervention. Transit providers carry traffic. RIPE maintains the registration. Domain and certificate services keep public endpoints usable. Payment, Peppol and bank interfaces support invoicing. Point-of-sale and supplier interfaces feed RestoCore. An advertised connection to Anthropic's Claude adds another external service with its own price and data-handling terms.
Each provider can be rational in isolation. Together they create a first-dollar problem. Registry and base hosting costs recur before the first customer uses a new module. Connector work must be maintained even if only one client depends on it. Hardware components may be bought before acceptance. Security obligations rise with every integration. If MH underprices the first customer on the assumption that ten more will follow, the suppliers have already earned their return while MH is still financing product development.
Datacenter United is a particularly revealing contrast. It markets a national network of Belgian facilities, carrier-neutral connectivity, direct cloud connections, certifications, monitoring, remote hands and high service levels. It can spread those capabilities across many customers and locations. MH can buy a portion of that scale. It cannot reproduce the economics with a /22 and a small balance sheet.
This does not make outsourcing a weakness. It makes supplier selection and pass-through pricing central. MH should buy resilience where it is cheaper than building it, disclose the boundary to customers and avoid guaranteeing more than the upstream contract supplies. A customer who requires a five-nines service should pay for the architecture and support required, not infer it from the existence of three name servers.
The downside sits with MH if contracts promise outcomes while supplier terms offer only inputs. It sits with the customer if MH disclaims continuity and the system is hard to replace. Good contracts align those risks. Weak contracts leave the owner making unpaid interventions to preserve reputation.
Contract durability is the only credible moat
At this scale, neither code volume nor address count is a moat. Code can become obsolete; addresses can be rented or transferred; competitors can offer similar feature lists. Durable value comes from contracts that are hard to displace for good reasons: deep process fit, reliable data migration, well-maintained interfaces, documented continuity and a price lower than the customer's cost of switching.
MH has ingredients for such contracts. Its software spans administrative and physical operations. RestoCore can hold recipe, supplier and stock logic that becomes more useful over time. Electronics can connect equipment that a generic SaaS product cannot see. Stable addresses can preserve external allow-lists. Local Dutch-language support and Belgian hosting can matter to customers with data-location or intervention requirements.
The danger is confusing lock-in with retention. A proprietary framework and development environment may make MH efficient, but they can also make the customer dependent on one supplier. If documentation, source access, data export, recovery procedures and succession are unclear, the customer will discount the contract or demand a lower price. Larger buyers may reject the risk entirely.
A small company can answer this without giving away its intellectual property. It can provide tested exports, a recovery runbook, named backup support, code escrow for defined failure cases, clear ownership of customer data and a transition service. It can publish service boundaries and measure restoration. Those commitments turn founder dependence from an unpriced fear into a priced term.
Contract length then becomes meaningful only with margin. A three-year agreement that underprices support destroys more value than a short project. The metric that matters is recurring gross profit after upstreams and expected support, not the logo count or contract face value. MH needs fewer good contracts, not more obligations.
Customer concentration remains the hidden balance-sheet risk
MH does not publish customer counts, references, sector mix or contract concentration. The accounts are too small and volatile to infer a broad base. A swing from negative gross margin of roughly EUR 38,000 in 2022 to positive EUR 14,600 in 2023 and back below zero in 2024 is consistent with project timing, concentration or both, though it does not prove either.
Founder-led custom suppliers often face a paradox. One large client provides efficient selling and valuable domain knowledge. It also gains bargaining power, absorbs delivery capacity and can remove the supplier's margin with a delayed project or renewal. Diversification reduces that exposure but increases sales work, onboarding and connector variety. At MH's visible scale, ten unrelated custom clients may be less manageable than three clients on a common product.
The answer is cohort discipline. Hospitality customers using the same RestoCore modules should share a release path and connector catalogue. Industrial electronics customers should share hardware and communication foundations. Generic CRM or invoicing work should be accepted only when it strengthens those foundations or carries a price high enough to justify divergence.
The public site claims experience across hospitality, industry, retail and construction. That can reassure a prospect that MH understands varied operations. Economically, it risks spreading one person's attention across sectors with different regulations, buying cycles and support windows. Broad sector language is marketing until revenue mix, reusable modules and service capacity show that the breadth lowers risk.
Cash collection is as important as concentration. With about EUR 10,850 cash and negative working-capital indicators in the public summary, MH has limited room to finance a customer's implementation. Milestone billing, deposits, automatic recurring payment and suspension rights are not aggressive terms; they are the mechanism that keeps a small supplier from becoming an involuntary lender.
Standard SaaS turns broad functionality into a commodity
The broad software market is hostile to undifferentiated custom work. Statbel's 2025 survey found that three in five Belgian enterprises use enterprise resource planning software. Basic CRM, invoicing, products, inventory, purchasing and reporting are not novel categories. Buyers can compare mature suites, implementation partners and app ecosystems.
Odoo illustrates the pressure. Its public offer includes one application free for unlimited users, then all-app plans priced per user. The custom tier adds external interfaces, multi-company support and several hosting choices. Its fees do not include every implementation or custom-code maintenance cost, but the buyer receives a large application set, regular upgrades, cloud monitoring, backups and support. MH must beat the total fit and ownership cost, not just the licence line.
Lightspeed is more direct competition for hospitality attention. In Belgium it advertises restaurant plans at EUR 89, EUR 159 and EUR 249 a month, extra licences at EUR 49, around-the-clock phone and chat support, inventory features, interfaces and a partner marketplace. It says it served roughly 144,000 customer locations at 31 March 2025. Some features cost extra, and a restaurant may still need the deeper recipe and supplier logic MH offers. But Lightspeed sets expectations for price transparency, support coverage, hardware integration and proof of adoption.
Cloud infrastructure applies the same pressure below the application. A customer can buy compute, managed databases, monitoring, backups and security tools from global providers. A local supplier can add accountability and data location, but it cannot claim that owning IPv4 addresses alone produces a cheaper or more resilient stack.
MH's realistic alternative is not to replicate these vendors. It is to attach itself where their standardisation stops: an unusual recipe structure, a supplier interface, a physical controller, a legacy serial device, a local reporting requirement or a workflow whose redesign would cost more than adaptation. Every feature outside that boundary should face a buy-versus-build test.
Local customisation can still win at the edge
Scale is powerful but not universal. Large vendors optimise for repeatable demand. Their support may understand the standard product but not the customer's machine, till, supplier contract or staff routine. Their road map can remove a feature. Their acquisition economics may not justify a small Belgian niche. This leaves room for MH.
The highest-value offer would combine three elements already visible in its work. First, a stable application foundation handles permissions, multilingual data, invoicing and reporting. Second, a vertical layer handles hospitality or industrial semantics. Third, an integration layer connects equipment and external systems. The address estate supports hosting and stable access but stays invisible to the buyer unless it improves continuity.
That is differentiated demand. A restaurant should not choose MH because it has a /22. It should choose MH because RestoCore measures recipe cost and stock correctly across an existing till, exports data cleanly and receives competent local support. A manufacturer should not choose MH because it belongs to RIPE NCC. It should choose MH because a controller, serial network and administrative application work as one supported product.
This also clarifies pricing. MH should charge for avoided operational cost, not for the number of screens or development hours. A stock system that prevents waste or catches margin erosion can justify a fee tied to the value at risk. A control system that avoids downtime can justify a continuity premium. In both cases, the company must prove outcomes and standardise support enough to retain part of the value.
The address resources then become an enabling asset rather than the thesis. They can support stable service endpoints, customer allow-lists and migration options. That is enough. Trying to build a general hosting or telecom offer around them would place MH against suppliers with far more capital, automation and purchasing power.
Regulation creates leads and obligations at the same time
Belgium's mandatory structured business-to-business electronic invoicing from 1 January 2026 is a clear demand catalyst. The federal finance authority specifies invoices compliant with the European standard and exchange through the Peppol framework. MH already advertises Peppol in its invoicing module. Existing custom-software customers may need upgrades, and small firms may seek help connecting operational data to compliant invoices.
The opportunity is not automatically high margin. Many accounting and enterprise suites include compliant invoicing, and the public authority warns that listing an application does not amount to approval or certification. MH must sell integration and workflow value around Peppol, not the transport standard itself.
Cybersecurity scope is more nuanced. Belgium's NIS2 law generally applies by sector and company size, with important exceptions. Public electronic communications providers and certain domain or trust services can fall within scope regardless of size. The RIPE membership record alone does not establish that MH provides a public communications service, and the BIPT list produced no match. If MH remains a private technology and hosting supplier, direct classification may differ; its larger customers can still impose supply-chain controls by contract.
Data protection applies at the service boundary regardless of marketing category. CRM, invoicing, restaurant loyalty and employee permissions can involve personal data. Customer contracts need clear roles, subprocessors, retention, access control, incident handling and deletion. Hardware that connects to operational networks adds patching and lifecycle expectations. These obligations consume engineering time even when no new feature is sold.
Regulation therefore favours suppliers that can spread compliance work. For MH, the winning response is narrow reuse: one Peppol integration, one security baseline, one data-processing schedule and one recovery practice applied across a coherent customer group. Bespoke compliance for every client would deepen the scale disadvantage.
Geopolitical locality helps only when it is contractual
MH can plausibly offer a Belgian-control story. The company is Belgian, its visible IPv4 route is originated by a Belgian data-centre network, and the upstream advertises local facilities, data-location benefits and national redundancy. European customers increasingly examine where operational data sits, which law applies and whether support is reachable locally.
But locality is not sovereignty by assertion. The public evidence does not identify MH's server sites, backup countries, subprocessors or disaster-recovery topology. Its software advertises a connection to a US AI provider. Point-of-sale, payment and cloud interfaces may cross jurisdictions. Electronic components are exposed to global supply chains. A Belgian address and RIPE country code do not resolve those dependencies.
Geopolitical risk is also commercial. A global platform can change price, restrict a service, alter data terms or prioritise larger markets. A component shortage can make an electronics design costly to reproduce. Energy and data-centre pricing can rise. At the same time, buying local resilience from a scaled Belgian upstream may be less risky than operating equipment alone.
The advantage becomes real only when MH maps dependencies and makes enforceable choices: specified hosting and backup locations, approved subprocessors, export and exit rights, component substitution plans, and a service design that can move the IPv4 block or applications if an upstream relationship ends. Customers who value those terms can pay for them. Customers who do not will choose on price and features, where global scale usually wins.
Public signals show technical depth and commercial underdevelopment
Non-official signals must be treated carefully, but they help frame what to verify. MH has a long technical trace: the company dates to 2007, its RIPE contact history reaches back further, the resource allocations have existed since 2013, and public technical forum records show electronics activity. This is not a newly assembled sales page with no operating history.
The current website is also thin as commercial proof. It publishes no named customer, case study, uptime history, security page, price, support window, partner list or deployment count. Two product pages contained visible editing residue or repeated copy when accessed. Those details do not prove poor engineering. They signal that the public offer has not received the same standardisation that a buyer expects from a mature product vendor.
Search discoverability is weak. The RestoCore name competes with unrelated restaurant education and reservation sites, while older local directories appear to have interpreted it as a restaurant at MH's former address. There were no visible customer reviews tied clearly to the Belgian software offer in the searches conducted. Again, that is a market signal, not evidence that the software has no customers.
The gap matters because a founder-led supplier needs trust before a buyer accepts continuity risk. Technical specificity helps; proof closes the sale. A short case showing the old process, integration, measured saving, support arrangement and customer reference would be more valuable than another module page. Publishing a service boundary and recovery commitment would do more for infrastructure credibility than emphasising cloud language.
Commercial underdevelopment can be fixed more cheaply than network scale. The question is whether management chooses focus. A broad website can attract varied enquiries but also encourages low-fit work. A narrow proposition with proof may produce fewer leads and better gross profit.
Capital allocation should favour proof, not breadth
The 2024 balance sheet does not support a speculative expansion into carrier infrastructure, a large employed sales force or multiple new products. The company needs proof that existing assets can earn their cost. That implies a cold sequence of decisions.
First, preserve the address estate while measuring its use. Identify which paying services require stable IPv4, what each address supports, the cost of the upstream arrangement and the recovery plan. Enable IPv6 only against a funded service or a clear reduction in future cost, but stop treating the dormant allocation as evidence of capability.
Second, choose the highest-margin repeatable offer. RestoCore has a defined buyer and operational pain, but the company should know whether hospitality credit and support risk outweigh reuse. The software-electronics combination may offer fewer customers and higher differentiation. Generic business software should not absorb development unless it converts into modules used repeatedly.
Third, repair the contract and proof layer before adding features. Price onboarding, recurring service, changes and continuity separately. Collect cash before committing external cost. Publish references and service boundaries. Establish backup delivery capacity and tested export. Measure recurring gross profit after every supplier and expected support hour.
Fourth, use outside scale deliberately. Datacenter United can provide facilities and reachability; established software can provide commodity functions; till vendors can process payments. MH should integrate these inputs where they are economical and own only the layer for which customers pay a premium. Strategy without that resource allocation is marketing.
Finally, restore the balance sheet. Positive equity and a larger cash buffer are not cosmetic. They give a small supplier room to replace hardware, respond to incidents and refuse bad projects. Selling more work at negative value added would make the company busier and weaker.
Five facts would change the judgment
The first is contract quality. Evidence of material annual recurring revenue, high gross retention and support-adjusted gross margin would show that the software and infrastructure footprint earns rent. A long contract is insufficient without its contribution after upstream and service costs.
The second is customer diversification. A base of customers using the same modules, with no single account able to remove a destabilising share of gross profit, would reduce the concentration discount. Named references would also clarify which of MH's many capabilities buyers actually value.
The third is continuity depth. Independent backups, a second service site, tested restoration, documented escalation and delivery capacity beyond one person would make the local-control proposition credible. An independently routed second path or a clearly tested prefix-move procedure would improve the network case, though MH does not need its own autonomous system if the contract outcome is strong.
The fourth is dual-stack use. A visible, secure IPv6 deployment tied to real services would show that the /32 is an operating asset rather than dormant administration. It would also reduce the impression that the infrastructure proposition depends entirely on IPv4 scarcity.
The fifth is financial repair. Positive gross margin across several years, positive operating cash flow, restored equity and a cash buffer sized to service obligations would demonstrate value creation. Revenue growth without those outcomes would not change the conclusion.
There are also facts that would worsen it: further debt growth, loss of the upstream route, a major customer departure, unresolved product-support commitments, or address monetisation used merely to fund recurring losses. The public evidence does not establish any of those events. They are the watchpoints implied by the current structure.
The conclusion: an infrastructure price-taker until contracts prove otherwise
MH is not an empty shell and its resources are not decorative. The company has operated for nineteen years, holds a cleanly registered IPv4 /22 and IPv6 /32, uses the IPv4 space for visible services, maintains proprietary software foundations and describes credible electronics work. Those assets give a small Belgian supplier more control than an ordinary consultancy renting every element by the month.
They do not yet produce cloud-scale economics. The IPv4 route depends on Datacenter United's autonomous system. The IPv6 allocation is not publicly routed. The company publishes no independent service depth, telecom notification, price architecture, customer proof or recurring-revenue measure. The latest available financial summary shows negative value added, negative EBITDA, a EUR 20,000 loss, debt above assets and limited cash.
Who pays? Customers pay when adaptation solves a problem that standard software cannot. Who benefits? MH benefits from reused code and stable service addressing, while upstream vendors receive recurring fees. Who carries the downside? The owner carries financing and capacity risk; customers carry continuity and switching risk unless contracts explicitly move it back to MH. That is not yet a balanced allocation.
The explicit judgment is therefore unfavourable but not terminal. MH is currently an infrastructure price-taker with useful technical options, not a differentiated infrastructure business. Its best route to value is to narrow the commercial offer around software-plus-electronics problems, make continuity contractual, charge fully for custom obligations and prove recurring gross profit. If those contracts emerge, the resource-holder footprint will strengthen a defensible niche. Without them, the addresses are an asset beside the business, not evidence that the business has escaped the margin risk below cloud scale.