The Numbers Behind Gases for Glass

This is a narrow, technical slice of the industrial gases business, and the first thing to know is that nobody agrees on how big it is. Published 2025 estimates for industrial gases sold into glass manufacturing run from roughly $4.8 billion to $8.4 billion. One widely quoted outlook puts the category at $5.3 billion in 2025, reaching $12 billion by 2034 on a 9.5% CAGR (Research and Markets, 2025). Another sizes the same category at $8.38 billion in 2025 growing to $11.37 billion by 2035, a far gentler 3.1% CAGR (Market Research Future, 2025). A third tracks the category to $5.99 billion by 2028 at 9.3% (The Business Research Company, 2025).

That spread is not a nuisance to be averaged away. It is a planning input. The high estimates count the total value of gas consumed by glassmakers, including gas produced inside on-site plants owned by the glassmaker. The low estimates count merchant sales only. If you are writing a plan for a merchant supply business, build your addressable market from the lower number, state which definition you used, and say so in one sentence. Credit committees spot an unqualified market figure faster than anything else in a plan, and in this category you can show your working in two lines and look like the only person in the room who read the source notes.

Where the money sits by gas and by glass type

Oxygen is the largest single product segment at over 40% of the category, because it feeds combustion in melting furnaces (Market Research Future, 2025). Nitrogen follows, used to inert the process, prevent oxidation and control the atmosphere during forming and surface treatment. Hydrogen, usually as a nitrogen and hydrogen blend, maintains the reducing atmosphere over the molten tin bath in float glass so the tin does not oxidise and mark the ribbon. Argon appears in insulating glass unit filling and some specialty melts, and helium in leak detection and coating lines.

On the demand side, the category splits across float glass, container glass and specialty or technical glass. Float lines are the biggest single consumers per site and the hardest to win, because they are almost always on long-term contracts with one of the global majors. Container glass plants are more numerous and more winnable. Specialty glass, including low-iron solar glass and coated energy-efficient glazing, is the fastest growing demand pool and the one where applications knowledge is worth more than tanker capacity.

Who you are actually competing with

Three companies hold most of this market. Published share estimates put Linde at roughly 30%, Air Liquide at roughly 24% and Air Products at roughly 16% of global industrial gases (von Arvy analysis, 2025), with Messer Group named as the third global competitor in regional segments alongside Air Liquide and Linde. The UK is more concentrated still: Linde holds about 60% of the UK gases market, Air Products 22%, Air Liquide 9% and Nippon Gases UK 6% (gasworld UK Country Report).

That 6% figure is the single most useful number on this page for a new entrant. Nippon Gases is a multinational with its own production assets and it holds 6% of a mature national market. Your plan should not project displacing an incumbent on price. The majors own the air separation units, so their marginal cost of oxygen is structurally below your purchase cost. What they do not have is a senior combustion engineer who will spend three days on a customer's furnace crown taking readings. That asymmetry is the entire basis of a credible entry strategy, and it needs to be stated plainly in the competitive section rather than buried in a SWOT grid.

The decarbonisation driver, in concrete terms

Glassmakers are rebuilding furnaces to cut carbon, and every one of those rebuilds is a procurement event. Pilkington UK ran the first hydrogen firing trial in a glass furnace at its St Helens site in 2021, initially firing one section on 100% hydrogen to replace about 20% of natural gas volume, then admixing hydrogen across all burners at a total share capped around 15%. The same furnace later ran on 100% bio-oil for four days with no effect on glass quality (British Glass). Encirc is building an ultra-low-carbon hybrid furnace at Elton in Cheshire for production from 2027, targeting a 90% emissions reduction on green electricity and low-carbon hydrogen supplied through the HyNet North West cluster, in partnership with Diageo (Glass International). Ardagh Glass modernised the W1 furnace at its Doncaster plant, an investment estimated to cut 20,000 tonnes of CO2 over a minimum 15-year campaign (Glass International).

Name those projects in your plan. A funder reading an industrial gases plan wants evidence that demand is tied to dated, capital-committed events rather than a generic growth curve. Three named furnace projects with timelines does that work in a paragraph.

Three Ways Into This Market

Most plans in this category fail because they describe "an industrial gases business" without choosing a model. The three routes have different capital requirements, different contract lengths and completely different risk profiles. Pick one as your core and name the others as stage-two options.

The on-site model is where the economics are most attractive and the financing is hardest. VPSA oxygen production has been quoted at around $0.03 per normal cubic metre against roughly $0.10 for purchased liquid oxygen, with reported savings of 40% to 60% over liquid supply arrangements (PKU Pioneer / VPSATech). Those are vendor figures and should be treated as a ceiling rather than a forecast, but the direction is not in dispute: above roughly 20 to 30 tonnes of oxygen per day, generating beats trucking. Below that, liquid supply wins on capital efficiency.

For a first-time founder without a balance sheet, the realistic sequence is model one into model two. Sell the applications engineering first, use the furnace survey data it generates to size the gas load accurately, then bring a lender a distribution business with real offtake evidence attached. That is a materially stronger loan application than a cold projection, and it is the sequence we build into the plans we write for this sector. If your concept sits closer to building the equipment than supplying the gas, the cryogenic tanks business plan template covers that angle instead.

What It Costs to Get Started

The figures below are for a merchant supply business entering through packaged gas and applications engineering, scaling into bulk cryogenic distribution. Total range: $185,000 to $2.4 million in the US, or £145,000 to £1.9 million in the UK. The low end is a two-person operation with a leased depot, a modest cylinder fleet and a distributor offtake agreement. The high end funds a tanker, a 3,000-cylinder fleet, a compliant filling facility and six months of gas purchases.

Line-by-line capital requirement

• Cryogenic or microbulk delivery vehicle: $120,000–$380,000 (£95,000–£300,000). A used 3,000 to 5,000 gallon liquid oxygen or nitrogen tanker sits at the bottom of that band. New microbulk vehicles with onboard pumps and metering sit at the top.
• Cylinder and dewar fleet: $90,000–$260,000 (£70,000–£205,000) for 1,500 to 3,000 units including specialty mix cylinders and 180-litre liquid dewars. Treat this as a depreciating asset with an annual loss rate, not a one-off purchase.
• Depot lease, concrete pad, bunding and fire separation: $40,000–$150,000 (£32,000–£120,000). Separation distances for bulk oxygen under NFPA 55 in the US, and the siting requirements in BCGA Code of Practice CP36 in the UK, are what drive the footprint and therefore the rent.
• Filling manifold, high-pressure pumps, purity analysers and calibration bench: $75,000–$300,000 (£60,000–£240,000). Skip this entirely in year one if you fill through a partner and resell. It is the single largest capital decision in the model.
• Regulatory setup: $6,000–$28,000 (£5,000–£22,000). PHMSA hazmat registration, a cylinder requalification identification number if you test in-house, a Dangerous Goods Safety Adviser retainer in the UK, and ADR or hazmat-endorsed driver training.
• Insurance: $35,000–$95,000 per year (£26,000–£72,000). Product liability, environmental impairment, commercial auto and workers compensation. Oxidiser handling moves you into a higher premium band than general chemical distribution.
• Tank telemetry, ERP and route optimisation: $18,000–$60,000 (£14,000–£48,000). Remote tank level monitoring is not optional. It is how you convert delivery from a schedule into a demand signal, and it is the difference between 70% and 92% tanker utilisation.
• Working capital, four to six months of gas purchases: $120,000–$500,000 (£95,000–£395,000). Your supplier wants 30-day terms. Your glass customers will push for 60 or 75. That gap is the number that kills gas distributors, and it belongs in the cash flow forecast before anything else.

The cost nobody models until it hurts

Liquid oxygen and nitrogen boil off. A vacuum-insulated vessel loses product continuously, typically a fraction of a percent per day, and more if the vacuum degrades or the vessel sits part-full. Every tonne that boils off was bought and will not be invoiced. Model boil-off as a cost of goods line with an explicit assumed percentage, size your customer vessels to the actual draw rather than to the biggest tank the customer will accept, and include a vacuum integrity check in the maintenance schedule. Plans that omit this read as written by someone who has never operated a cryogenic asset, and experienced lenders in industrial sectors notice.

Funding Routes and the SBA Reality Check

In the US, the primary route is an SBA 7(a) loan, capped at $5 million with terms to 25 years on real estate and 10 years on equipment and working capital. SBA Express and Export Express caps are lower at $500,000 (US Small Business Administration). The size standard for NAICS 325120, Industrial Gas Manufacturing, is 1,200 employees averaged over the trailing 24 months (NAICS 325120), so size eligibility is never the constraint for a new entrant in this category.

Scale expectations matter more. Across all 7(a) lending the mean loan is about $523,600 but the median is $195,000, and the national average for FY2024 was roughly $443,000, down from $538,000 in FY2023 (SBA loan statistics compilation). Read that against the cost table above and the conclusion is uncomfortable but useful: a typical 7(a) facility funds the packaged and applications entry model comfortably, and funds roughly a third of a full bulk distribution launch. If your plan asks for $1.8 million against a median of $195,000, you are not making a normal request and the narrative has to earn the exception.

What makes this niche hard to underwrite, and how to fix it

Two structural problems. First, collateral quality. A cylinder fleet is portable, individually low-value and prone to shrinkage, so lenders discount it heavily. A tanker holds value better but is specialist and slow to resell. Second, revenue concentration. Six customers generating 90% of revenue is normal in this business and alarming to a credit analyst who normally sees hundreds of accounts.

Both are answerable. Bring signed heads of terms or letters of intent from at least two glass plants, with indicative volumes and a take-or-pay minimum stated. Show the furnace-level calculation behind each volume, not a market-share assumption. Include the contract outage clause language in an appendix. And show the applications-engineering revenue line separately, because it is asset-light, high-margin and demonstrates that you can earn without the gas volume arriving on schedule.

Outside the US

In the UK, the Start Up Loans scheme offers up to £25,000 per founder at 6% fixed with free mentoring, which in this sector funds the applications-consulting phase but not the assets. Realistic UK routes for the asset-heavy phase are asset finance against the tanker and vessels, the British Business Bank Growth Guarantee Scheme through a participating lender, and regional grant support where a furnace decarbonisation project is involved. Comparable programmes exist in Canada through BDC, in Australia through the major trading banks, and in the UAE through the Khalifa Fund. In every case, asset finance against the vehicle and vessels sits alongside the main facility rather than inside it, which improves the headline debt service coverage ratio.

For a broader view of funding structures across the whole gases sector, the industrial gases business plan template covers the wider category. Our Research + Content package builds the lender-facing version of the volume calculation described above.

How the Money Works

Revenue in this business comes from six places, and they have very different margin characteristics. A plan that models only the first one will look thin to anyone who knows the sector.

• Bulk liquid supply under take-or-pay: delivered oxygen in competitive US corridors runs roughly $95 to $145 per tonne against an ex-plant purchase cost of $62 to $85. Gross spread of 30% to 40% before distribution cost, which eats most of it.
• Vessel and vaporiser rental: $450 to $1,800 per month per customer installation. Small numbers, near-100% margin, and the stickiest revenue you will ever book because removing a vessel requires a crane and a shutdown.
• Packaged cylinders and microbulk: 38% to 52% gross margin on forming, coating, laboratory and maintenance gas. Lower volume, much better margin per cubic metre than bulk.
• Specialty gases and calibration mixes: helium for leak detection, certified mixes for coating line process control and emissions monitoring. Highest margin per unit in the portfolio.
• Applications engineering: oxy-fuel conversion studies, oxygen boosting trials, burner retrofit supervision, campaign-extension analysis. Day rates of $1,400 to $2,600. This is the wedge, not a sideline.
• Cylinder requalification and testing services: hydrostatic and visual requalification for your own fleet and third parties. Modest revenue, useful asset utilisation.

Sizing a glass customer: from furnace pull to invoice

The one calculation that separates a serious plan from a generic one is converting furnace pull into gas volume. Published figures put total oxygen demand for an oxy-fuel glass melting furnace between about 5,000 scfh (roughly 5 tonnes per day) and 300,000 scfh (roughly 300 tonnes per day) for furnaces pulling 10 to 600 tonnes of glass per day (US Patent 6,398,547). That implies an oxygen-to-glass mass ratio in the region of 0.25 to 0.5, varying with furnace type, cullet ratio, and whether the furnace is fully oxy-fired or only oxygen-boosted. Treat 0.30 as a planning midpoint and replace it with the customer's own fuel meter data before you sign anything. Labelled as an Avvale planning estimate, that ratio is defensible. Presented as a fact, it is a liability.

Why the applications work carries the pricing

You cannot win this market on price, so you have to be able to price on saved energy. The numbers are strong enough to support that conversation. Oxy-fuel firing has delivered fuel savings of 10% to 15% on campaign average for large container and float furnaces with efficient regenerators, and 30% to 50% on fibreglass furnaces with metallic recuperators, within an overall reported range of 5% to 50% against air-fired operation. Environmental results are more dramatic still: NOx down by up to 90%, carbon monoxide down by up to 96%, and particulates down by up to 30% (ACEEE, Oxy-Fuel Firing for the Glass Industry).

Energy used to produce the oxygen offsets part of the saving, which is exactly why the honest version of this pitch is more persuasive than the vendor version. Show the net position. A supplier who arrives with a furnace-specific net energy balance, including the power cost of oxygen production, is doing something the incumbent's account manager usually is not, and that is what justifies a delivered price above the cheapest quote on the table.

The equipment side of this conversation is well documented and worth naming in your plan. Air Liquide's ALGLASS burner family is designed as a non-water-cooled oxy-fuel burner for continuous melting furnaces (Air Liquide UK). Linde's OPTIFIRE JL burner, built on its Dilute Oxygen Combustion technology for ultra-low NOx, has been installed in over 50 furnaces (Linde OPTIFIRE JL datasheet). Knowing which burner platform a target furnace runs, and what it was commissioned to do, is the kind of detail that opens a technical conversation with a plant engineer.

Permits, Pressure Law and Driver Compliance

There is no single "industrial gas supplier licence" in any major market. You clear a stack of transport, pressure-equipment and site-safety requirements, and the sequencing matters because several of them gate each other. A plan that states a launch date without a permitting critical path is the fastest way to lose credibility with an industrial lender.

United States

• Hazardous Materials Regulations, 49 CFR parts 171 to 180 (PHMSA, US DOT). Cylinders filled with a Class 2 gas and offered for transport must comply with the HMR, and may not be offered or transported unless maintained and requalified under it. 49 CFR 173.301 sets the general requirements for shipping compressed gases in cylinders and UN pressure receptacles.
• PHMSA hazmat registration. Annual, $25 to roughly $3,000 depending on company size. Four to ten weeks. A prerequisite for offering hazmat in commerce, not an afterthought.
• Cylinder requalification identification number (RIN) if you hydrostatically test in-house. PHMSA has been updating the HMR to incorporate or refresh references to Compressed Gas Association publications, amend filling requirements for compressed and liquefied gases, and clarify cylinder manufacture and requalification rules (Federal Register, 2020). Track those amendments; your SOPs reference them by number.
• OSHA compressed gas standards. Handling, storage and use of all compressed gas cylinders must follow CGA Pamphlet P-1. OSHA expects cylinders stored so they cannot be damaged or knocked over, valves and caps protected, and condition controlled by inspection and testing (OSHA interpretation).
• CDL with hazardous materials endorsement for every driver, including the TSA security threat assessment. $100 to $450 per driver, four to eight weeks. Build driver recruitment into the critical path, because the endorsement is not instant and a tanker without a qualified driver is parked capital.
• Local fire marshal permit and NFPA 55 compliant siting. $500 to $6,000, six to sixteen weeks. Separation distances for bulk oxygen systems and exposure protection determine the depot footprint, so settle this before signing a lease.

United Kingdom

• Pressure Systems Safety Regulations 2000. Static cryogenic tanks are pressure systems. Each installation needs a written scheme of examination, periodic formal examination, routine maintenance and a designated competent person. Budget £1,200 to £6,000 per written scheme plus examination fees, four to ten weeks per installation.
• Pressure Equipment (Safety) Regulations 2016. New cryogenic tanks and their associated systems must be manufactured and put into service under PE(S)R. This is handled at manufacture, but your plan should say who holds the conformity documentation.
• Carriage of Dangerous Goods and Use of Transportable Pressure Equipment Regulations. Covers transportable tanks and cylinders. You must appoint a Dangerous Goods Safety Adviser (£1,500 to £4,000 retainer) and train drivers to ADR (£600 to £1,200 each).
• BCGA Codes of Practice. CP36 for cryogenic liquid storage at users' premises, CP39 for in-service requirements of pressure equipment, CP27 for transportable vacuum-insulated vessels. These are the documents an HSE inspector will reference. CP36 is explicit that installations must be sited to minimise risk to personnel, local population and property, with regard to nearby hazardous processes, and that foundations are agreed with the tank supplier while installation responsibility usually rests with the site owner or operator (BCGA CP36).
• COMAH screening. Large oxidiser inventories can bring a site into scope of the Control of Major Accident Hazards Regulations. Check the threshold against your planned liquid oxygen inventory before you design the depot, not after.

India, and the wider picture

India is a live market for this niche because of solar glass capacity growth, and its licensing regime is unusually specific. The Petroleum and Explosives Safety Organisation administers the Gas Cylinders Rules 2016, under which no person may fill a cylinder with compressed gas, and no filled cylinder may be possessed, except under a licence granted under those rules. Rule 43 establishes that filling and possession licensing requirement subject to stated exceptions (PESO FAQ on the Gas Cylinders Rules). The SMPV(U) Rules 2016 separately govern design, manufacture, installation and use of static and mobile pressure vessels (PESO, SMPV(U) Rules 2016).

One PESO distinction is worth building a strategy around. A PSA or VPSA plant that supplies generated gas directly, with no cylinder filling, does not require a licence under the rules PESO administers. Attach a compressor and a filling point and the licensing obligation applies. For an India entry, that single boundary can determine whether you launch in four months or fourteen.

In the EU, the equivalent stack is ADR for road carriage, the Pressure Equipment Directive 2014/68/EU for the vessels, and Seveso III thresholds for larger oxidiser inventories. The pattern is consistent across jurisdictions: transport rules, pressure-equipment conformity, and site inventory thresholds. Write your compliance section around those three headings and it will travel.

Six Mistakes That Sink Gas Supply Plans

These are the recurring failures we see in plans for this niche, in rough order of how much damage they do.

1. Pricing gas with no electricity pass-through

Electricity is the dominant input cost in air separation and in VPSA generation. A fixed price per tonne over a five-year contract transfers all power-price risk to you, and in a volatile power market that can erase the entire gross spread modelled earlier on this page. Every supply agreement needs an indexed pass-through tied to a named published power index, plus a defined review mechanism. State the index by name in the plan.

2. Building the plan around oxygen only

Oxygen is the biggest volume and the thinnest margin. The nitrogen and hydrogen used over the float glass tin bath, and nitrogen in forming and surface treatment, are continuous, interruption-sensitive and better priced per cubic metre. A plan that forecasts oxygen tonnes and treats everything else as "other revenue" has mis-weighted its own business.

3. No take-or-pay minimum and no cold-repair clause

Glass furnaces are rebuilt on campaigns measured in years, and a cold repair takes the furnace out of production for months. Without a take-or-pay floor you have no revenue during that window. Without an explicit outage clause you may have a dispute about whether the floor applies. Typical structures run three to seven years with a minimum around 80% to 85% of forecast monthly volume. Put the clause language in an appendix.

4. Modelling customer count instead of furnace count

One 600 tonne per day float line can consume as much oxygen as sixty small industrial accounts. Counting customers makes a concentrated book look diversified. Build the revenue model furnace by furnace, name each one, state its pull rate and firing configuration, and show the concentration honestly with a mitigation plan beside it. Lenders discount hidden concentration far more harshly than disclosed concentration.

5. Promising a launch date the regulator will not meet

The PSSR written scheme, NFPA 55 siting approval, PESO licensing and hazmat-endorsed driver recruitment all run on external clocks. Sequence them in a Gantt chart with the dependencies visible, and set the revenue start date after the longest chain, not after the shortest.

6. Treating the cylinder fleet as an expense

Cylinders are a depreciating asset with a real annual loss and shrinkage rate from customer retention, theft and scrappage. Carry them on the balance sheet, assume a loss percentage, fund replacement annually, and run a cylinder tracking system from day one. A fleet with no audit trail is also a compliance problem, because requalification dates become unprovable.

Terms Your Lender Expects You to Use Correctly

This sector has its own vocabulary, and using it precisely is part of how a plan signals competence. These eight terms appear in almost every conversation with a glass plant engineer.

• Pull rate. The tonnes of glass a furnace produces per day. Every gas volume calculation starts here, and it is the number to ask for first.
• Oxy-fuel firing. Burning fuel in near-pure oxygen rather than air, removing the nitrogen that would otherwise be heated and vented. The source of both the fuel saving and the NOx reduction.
• Oxygen boosting or enrichment. Adding oxygen to an air-fired furnace without full conversion. Lower capital, smaller saving, much easier first sale. Often the right opening offer.
• Regenerator and recuperator. Two heat recovery designs on air-fired furnaces. The distinction matters commercially: furnaces with efficient regenerators show smaller oxy-fuel savings than those with metallic recuperators, which is why fibreglass conversions show the biggest numbers.
• Cullet. Recycled glass charged with the batch. A higher cullet ratio reduces melting energy and therefore reduces oxygen demand per tonne, so the ratio belongs in the volume calculation.
• Tin bath. The molten tin on which float glass is formed. Protected by a nitrogen and hydrogen reducing atmosphere. A continuous, non-interruptible gas load.
• ASU and VPSA. Air separation unit (cryogenic, produces liquid product and high purity) and vacuum pressure swing adsorption (non-cryogenic, gaseous oxygen at around 90% to 93% purity). Glass melting tolerates VPSA purity, which is why on-site generation works here.
• Take-or-pay and facility fee. The minimum volume a customer pays for whether consumed or not, and the fixed monthly charge covering an on-site plant's capital recovery. Together they are what makes this business financeable.

Sample Business Plan Extract

An extract from an industrial gases supply plan written by our team, so you can see the level of specificity we work to:

What You Get in the Template

Every Avvale business plan template is pre-structured for the sector, with prompts that ask the questions a lender in that sector will ask. For this niche, that means the industry analysis section prompts you for furnace-level demand rather than market-share percentages.

• Executive Summary — the business in a page, built to hold a credit analyst's attention through the first sixty seconds
• Company Overview — legal structure, ownership, depot location and the founding story that explains your technical credibility
• Industry Analysis — market sizing with the definition stated, named decarbonisation projects as demand evidence, and the concentration picture
• Customer Analysis — plant-by-plant target list with pull rate, firing configuration and estimated gas intensity
• Competitor Analysis — incumbent mapping, structural cost disadvantage acknowledged, and the service asymmetry you are selling instead
• Marketing Plan — how a technical sale into a plant engineering team actually works, and why it is not a marketing funnel
• Operations Plan — depot layout against separation requirements, delivery routing, telemetry, cylinder tracking and the compliance calendar
• Management Team — founder credentials, the competent person appointment, DGSA or hazmat lead, and planned key hires

The optional Financial Forecast add-on, included in our $300 / £250 and $1,000 / £800 packages, provides a 5-year Excel model with income statement, cash flow, balance sheet, break-even analysis and startup capital schedule. For this sector we add a gas volume build-up tab so every revenue line traces back to a named furnace. If you are planning the manufacturing end rather than the supply end, see the glass manufacturer business plan template.

Questions Buyers Actually Ask