SCX.ai IPO: A$40 Million to Build Australia’s AI Compute Challenger

Noel Ong
2 days ago
15 min read

SCX.ai is arriving on the ASX today with an operational ten-rack AI inference node, A$5.35 million of contracted annual recurring revenue and a fully underwritten A$40 million raising. The central question is whether the company can convert a partner-supported launch into a diversified, high-utilisation infrastructure business before technology and competition move again.

SCX.ai Holdings Limited is listing on the Australian Securities Exchange today under the code SCX, giving investors exposure to an Australian-based artificial-intelligence infrastructure business focused on inference rather than model training.

The company offered 133,333,334 new shares at A$0.30 each to raise A$40 million before costs. With convertible notes also converting at listing, SCX expects to have 250,124,413 shares on issue and a market capitalisation of approximately A$75.0 million at the offer price.

SCX.ai IPO owns specialised SambaNova inference hardware installed inside the Equinix SY5 data centre in Alexandria, Sydney. It combines that hardware with its own software and customer service layer to sell token packages, dedicated computing capacity, managed AI services, and enterprise enablement.

The investment proposition is easy to understand. Australian enterprises and government agencies are expected to use more AI. Some workloads require Australian data residency and operational control, and inference should become a larger part of AI infrastructure spending as applications move from testing to production.

At a Glance

IPO detail	Prospectus position
ASX code	SCX
Offer price	A$0.30 per share
Gross IPO proceeds	A$40.0 million
New IPO shares	133,333,334
Convertible-note conversion shares	16,791,079
Total shares at listing	250,124,413
Indicative market capitalisation	A$75.04 million
Pro-forma net cash	A$29.03 million
Prospectus enterprise value	A$46.01 million
Options and adviser options	11,700,000
FY26 pro-forma forecast revenue	A$6.04 million
Contracted ARR at May 2026	A$5.35 million
Unbilled total contract value	A$14.61 million
Expected free float	56.7%
IPO underwritten	Yes - fully underwritten
Current anticipated quotation	21 August 2026 at 12:00 pm AEST

The 60-Second Pitch - SCX.AI IPO

SCX.ai IPO proposition rests on three ideas.

The first is market timing. AI expenditure is moving from training large models towards repeatedly running those models inside applications. That operating stage is inference. The prospectus's commissioned industry report estimates that inference could increase from approximately 49 per cent of AI infrastructure spending in 2025 to 69 per cent by 2030.

The second is sovereignty. Some Australian government, defence, healthcare, financial-services and enterprise workloads cannot be treated like ordinary public-cloud traffic. Customers may require local processing, stronger governance and more control over where prompts, documents and model outputs travel.

The third is infrastructure economics. SCX is using purpose-built SambaNova ASIC hardware rather than general-purpose GPU infrastructure. The prospectus reports material performance-per-watt and speed advantages for selected stable inference workloads. If those advantages translate into lower cost per token and higher rack utilisation, the model could generate operating leverage as demand grows.

The unanswered question is whether SCX can build a diversified commercial customer base quickly enough.

The current contracted metrics are dominated by SambaNova's three-year offtake agreement. Investors therefore need to distinguish between initial revenue visibility and independent market validation. Both matter, but they are not the same thing.

Figure 1: SCX History Overview (Source: ASX Announcement)

The Offer and What the Valuation Means

The IPO offered 133.33 million new shares at A$0.30 each. A separate conversion offer will issue approximately 16.79 million shares to holders of pre-IPO convertible notes.

Table 1: The offer particulars

Capital structure	Shares	Ownership at listing
Existing shareholders before note conversion	100,000,000	40.0%
Convertible-note conversion shares	16,791,079	6.7%
New IPO shareholders	133,333,334	53.3%
Total shares on listing	250,124,413	100.0%

At A$0.30 per share, SCX's indicative market capitalisation is A$75.04 million. The prospectus deducts A$29.03 million of pro-forma net cash to calculate an enterprise value of A$46.01 million.

That enterprise-value figure is more useful than market capitalisation alone because it separates the operating business from the cash raised. It implies that the market is placing an initial value of about A$46 million on SCX's installed hardware, software, contracts, team and growth platform after recognising the pro-forma cash balance.

There is another layer to that cash number. SCX has an A$11.34 million committed payment for the second tranche of ten SambaNova racks. If investors ring-fence that commitment, approximately A$17.69 million of the A$29.03 million pro-forma cash balance remains before other working-capital requirements. The expenditure is not lost - it is exchanged for additional hardware - but the cash is not fully discretionary.

The IPO also creates 11.7 million options and adviser options. The 4.7 million director and management options have an A$0.30 exercise price. The 7.0 million adviser options issued to Henslow and Canaccord have an A$0.33 exercise price and expire three years after listing. If all options were eventually exercised, the share count would rise by approximately 4.7 per cent from the listing base and the company would receive approximately A$3.72 million in exercise proceeds.

The Convertible Notes Deserve Attention

Between March and May 2026, SCX issued A$3.4 million of secured convertible notes for net cash proceeds of A$3.2 million. The money funded working capital, payments to SambaNova and IPO preparation costs.

The notes accrue interest at 2 per cent per month, with half payable in cash and half capitalised. On listing they convert at A$0.21 per share, representing a 30 per cent discount to the IPO price.

Bridge financing is not unusual for a company approaching an IPO, but these are expensive terms. The monthly interest rate, security and conversion discount reflect the risk accepted by noteholders and the company's need for capital before the public raising.

The note conversion will issue approximately 16.79 million shares, equal to 6.7 per cent of the listing capital. Directors, the joint lead managers and other noteholders participated in the notes. Henslow and Canaccord each received A$100,000 face value of notes as fees for managing the note offer.

The IPO underwriting itself carries cash selling, management and underwriting fees equal to 6 per cent of the gross proceeds, or A$2.4 million, plus the 7.0 million adviser options. Total estimated cash transaction costs are A$3.14 million. The adviser options were independently valued in the prospectus at approximately A$1.5 million.

This is part of the real cost of reaching the ASX and should be considered alongside the fully underwritten status of the A$40 million offer.

What SCX Actually Owns and Operates

SCX is not building a data centre and it is not developing a frontier foundation model.

Its model sits between those layers.

The company purchases specialised AI inference hardware, installs it in third-party Australian data centres and uses a proprietary software layer to turn computing capacity into services that customers can consume through application programming interfaces, dashboards and dedicated infrastructure arrangements.

The initial deployment consists of ten SambaNova SN40L racks at Equinix SY5 in Alexandria, Sydney. The FY26 forecast refers to 160 reconfigurable dataflow units across those ten racks. The data-centre allocation has approximately 120 kilowatts of capacity, while each rack typically consumes about 10 kilowatts.

The first node is therefore real and installed. This differentiates SCX from a concept-stage AI float that is raising money before acquiring its core infrastructure.

The second ten-rack tranche is also committed. Installation is planned to commence in November 2026, with service availability expected in early 2027. Beyond Sydney, SCX intends to add further Australian locations as demand develops.

The platform has four practical layers:

Infrastructure: Equinix SY5 provides the physical facility, power, cooling, security and connectivity.

Compute: SambaNova supplies the SN40L inference hardware and the software used to manage that hardware.

SCX software: SCX provides APIs, model access, routing, metering, dashboards, guardrails, storage, analytics and customer controls.

Enterprise services: The SCX team assists with deployment, workflow design, governance, compliance, fine-tuning and ongoing support.

SCX says it holds rights to three provisional patent applications relating to the software layer. The strategic value of the business will depend on how much differentiation and customer retention that layer creates beyond the underlying hardware.

Figure 2: Description of the SCX Platform (Source: ASX Prospectus)

Inference Is the Product - Not AI Training

Training is the expensive process of creating or materially developing an AI model. Inference is what happens afterwards: a trained model receives a prompt or request and produces an answer, classification, summary, image, transcription or software action.

Inference becomes recurring infrastructure demand when AI is embedded inside a business application. Every customer interaction, document query or automated workflow consumes tokens and computing capacity.

SCX is targeting that recurring stage through four revenue lines:

token-based inference subscriptions;
dedicated infrastructure leasing;
managed AI services; and
enterprise enablement and governance services.

The packaged subscriptions range from starter plans to growth and enterprise offerings. Dedicated arrangements allocate specific computing resources to a customer for a fixed term or minimum commitment.

This mix gives SCX the potential to combine recurring consumption revenue with larger contracted capacity agreements. It also means revenue can be volatile if customers take longer to move from trial activity into production workloads.

The Contracted Revenue Headline Needs Context

At May 2026 month-end, SCX reported:

A$5.35 million of contracted ARR;
A$14.61 million of unbilled total contract value; and
298 active platform users at 23 June 2026.

Those numbers show that the business has moved beyond a pre-revenue concept. They should not, however, be read without the concentration detail.

SambaNova accounts for A$14.4 million of the A$14.6 million unbilled total contract value. Its offtake agreement commenced on 1 April 2026, runs for an initial three years and requires a fixed monthly payment of US$302,553 for reserved compute capacity. The prospectus places the total contract value at A$15.2 million.

SambaNova is therefore doing three jobs at once:

selling SCX its specialised hardware;
providing the supporting software and managed services; and
purchasing capacity back from SCX under the offtake agreement.

This arrangement gives SCX immediate utilisation support and predictable monthly revenue while the external customer base develops. It also creates a circular concentration risk. If SambaNova suffers financial problems, changes strategy, is acquired, falls behind technologically or terminates an agreement following a breach, both SCX's revenue and its operating platform could be affected.

For FY26, SCX forecasts A$6.04 million of revenue. Approximately A$3.8 million, or 63 per cent, is underwritten by the SambaNova offtake agreement.

The remaining forecast still requires execution. A$2.13 million, or approximately 35 per cent of total FY26 revenue, is expected from new customer and subscription wins between June and December 2026. The forecast assumes 42 new customers or subscriptions, including two dedicated-rack subscriptions and 14 enterprise subscriptions.

At May month-end, SCX reported 69 identified and qualified opportunities, including 17 late-stage opportunities with service quotes submitted. That is useful pipeline evidence, but it is not contracted revenue.

The 298 active users also require care. The prospectus explicitly says the number includes both fee-paying users and non-paying users participating in pre-subscription trials. It should not be interpreted as 298 paying customers.

The cleanest post-listing commercial measure will be contracted ARR and total contract value excluding SambaNova, supported by disclosure on customer count, utilisation, retention and average revenue per customer.

The Financial Forecast Is a Seven-Month Execution Test

SCX's FY26 forecast combines five months of actual results to 31 May 2026 with seven months of forecast trading to 31 December 2026.

Table 2: Financial forecast

Pro-forma financial measure	FY25	FY26 forecast
Revenue	A$0.01m	A$6.04m
Cost of sales	Nil	A$2.52m
Implied gross profit	A$0.01m	A$3.52m
Implied gross margin	n.m.	58.3%
EBITDA	(A$2.21m)	(A$2.40m)
NPAT	(A$2.21m)	(A$3.55m)
Net operating cash flow	(A$0.51m)	(A$2.06m)
Capital expenditure cash flow	(A$1.33m)	(A$22.24m)

The implied 58.3 per cent gross margin is useful, but it is not yet a mature unit-economics result. SCX's cost of sales is largely fixed to installed capacity, including data-centre access, power, connectivity, SambaNova software and support. Margins should improve if more tokens are sold across the same racks, but they can deteriorate if utilisation remains low.

The company expects FY26 pro forma public-company costs of approximately A$1.2 million a year and executive-leadership cash costs of approximately A$1.9 million a year. Management has not provided a date by which it expects SCX to become profitable, saying there is no reasonable basis to do so.

The forecast sensitivity analysis shows how dependent the first year is on conversion timing. A one-month delay in new-customer conversion would reduce forecast EBITDA by A$353,000. The gain or loss of one dedicated-rack customer changes forecast EBITDA by approximately A$541,000.

This is why the next reporting periods matter more than the broad AI market-growth forecast. The business has to demonstrate that pipeline becomes contracted consumption before the cost base and next hardware tranche move too far ahead of demand.

Where the A$40 Million Goes

Table 3: Use of funds

Use of proceeds	Amount	Share of proceeds
AI hardware	A$19.3m	48.3%
Working capital	A$12.2m	30.5%
SambaNova managed AI services	A$5.4m	13.5%
Transaction costs	A$3.1m	7.7%
Total	A$40.0m	100.0%

The hardware allocation includes A$8.0 million to settle amounts payable on the first ten racks and A$11.3 million for the committed second ten-rack tranche.

Working capital includes A$3.9 million for personnel, A$2.8 million for sales and marketing, A$1.6 million for compliance, advisers and public-company costs, A$1.3 million for operating expenses associated with the next ten racks, and A$2.6 million for other working-capital items.

In total, A$24.7 million - approximately 61.8 per cent of the raise - is allocated to hardware and SambaNova-managed services. This confirms that SCX should not be valued as a conventional asset-light software company. It is a capital-intensive compute operator with a proprietary software and services layer.

The staged hardware model is sensible if the second node is matched to customer demand. The main execution risk is that capital is committed before sufficient independent utilisation has been secured.

Sovereign AI

SCX's central positioning is that customer workloads are processed on Australian infrastructure under Australian operational control.

That should appeal to organisations that care about data residency, regulatory alignment, security and the ability to deploy open-weight or private models without using a conventional public-cloud service.

The prospectus is also unusually clear about the limitation.

The Equinix facility is owned by a US-headquartered company, while SambaNova is a US hardware and software supplier. The prospectus says there can be no absolute assurance that customer data is entirely beyond foreign legal processes. US authorities could potentially seek access through foreign-owned infrastructure or technology suppliers, depending on the legal circumstances.

SCX may therefore provide stronger Australian control and residency than many public-cloud alternatives without delivering absolute technological or legal sovereignty.

That distinction matters most for defence, government and highly regulated customers - precisely the sectors central to the investment narrative. SCX will need to demonstrate that its architecture, contracts, governance and certifications satisfy the procurement standards of those customers in practice.

How SCX Should Be Compared

There is no clean ASX peer group for SCX.

Data-centre owners are primarily valued on land, powered capacity, occupancy and long-duration infrastructure economics. Traditional software-as-a-service companies are assessed on recurring revenue, retention, gross margins and customer-acquisition efficiency. Chip companies are valued on intellectual property, design wins and manufacturing economics.

SCX sits across these categories. It owns compute hardware but not the underlying data centre. It operates proprietary software but depends on SambaNova's architecture. It earns recurring and usage-based revenue, but the initial revenue base is dominated by an offtake partner.

The most useful operating measures are therefore:

contracted ARR excluding SambaNova;
independent enterprise and government customer count;
rack and token utilisation;
gross margin as utilisation increases;
customer retention and contract duration;
capital expenditure required for each increment of ARR;
hardware useful life and refresh costs; and
the timing of EBITDA and operating cash-flow break-even.

Applying a pure SaaS multiple would ignore the hardware intensity. Comparing SCX only with data-centre operators would ignore the software, customer layer and token economics. The investment case needs both lenses.

Who Is Running SCX?

David Keane, founder, Managing Director and Chief Executive Officer, previously co-founded Bigtincan Holdings and led that enterprise-software company through its ASX listing and eventual acquisition by Vector Capital in 2025. That experience is directly relevant: SCX needs to sell complex technology to enterprise customers while operating under public-market scrutiny.

Wayne Stevenson, independent non-executive chair, has banking, strategy and public-company experience, including previous roles at ANZ and as chair of QMS Media through its IPO. He was also a director of Bigtincan.

Thomas Amos, non-executive director, was the long-time chair of Bigtincan and is currently chair of Amber Technology. His background includes telecommunications, venture capital and public-company governance.

Penny Fowler AM, independent non-executive director, adds marketing, media, tourism and institutional board experience, including roles with Tourism Australia and Tech Mahindra.

Rahul Vaidya, Chief Financial Officer from 1 August 2026, brings 18 years of investment-banking, corporate-finance and technology-sector experience.

The board contains substantial Bigtincan continuity. That should support cohesion and knowledge of enterprise-software commercialisation, although investors should still assess whether SCX develops the deeper infrastructure, security and AI engineering capability required for a compute platform.

David Keane is expected to hold voting power of approximately 27.0 per cent on listing. Thomas Amos is expected to hold 8.6 per cent. Keane's holding is large enough to block resolutions requiring a 75 per cent majority, providing strong founder influence without majority control.

Approximately 42.4 per cent of the listing shares are expected to be subject to mandatory escrow, while the prospectus estimates a free float of 56.7 per cent.

One related-party matter is disclosed. Chief software engineer Robert Amos is the son of director Thomas Amos. He is employed on A$252,000 total fixed remuneration and has an interest in two million existing shares. The prospectus describes the terms as market-standard and arm's length.

Key Risks

Partner concentration: SambaNova is the core hardware supplier, software and support partner, and source of most contracted near-term revenue.

Customer diversification: Approximately A$14.4 million of A$14.6 million unbilled total contract value is attributable to SambaNova.

Forecast execution: A$2.13 million of FY26 forecast revenue depends on winning new customers or subscriptions after May 2026.

Rapid technological change: GPU inference, competing ASIC designs or more efficient models could reduce the economic advantage of the installed SN40L hardware.

Hardware refresh and capital intensity: Specialised compute hardware is expensive and may require faster replacement than the five-year accounting life assumed in the prospectus.

Single-site exposure: All current operating hardware is at Equinix SY5 in Sydney, creating concentration around power, cooling, access and connectivity.

Sovereignty limitations: The data centre owner and principal hardware supplier are US companies, potentially complicating claims of absolute sovereignty.

Competition: SCX competes with hyperscalers, specialist AI compute providers, telecommunications companies and other sovereign-cloud operators with greater resources and existing customer relationships.

Cybersecurity and data breaches: A security incident would directly damage the trust proposition underpinning the brand.

Procurement cycles: Government and regulated-enterprise sales can be slow, causing revenue to lag infrastructure expenditure.

Energy and foreign exchange: Power costs affect margins, while hardware, support costs and the SambaNova offtake are exposed to the Australian-dollar/US-dollar exchange rate.

No profitability timetable: SCX is loss-making and may require further funding for expansion beyond the first 20 racks.

Escrow release and dilution: Approximately 42.4 per cent of shares are expected to be escrowed, while 11.7 million options create potential future dilution.

Milestones That Will Test the IPO Thesis

The most useful post-listing milestones are specific and measurable.

Delivery of the A$6.04 million FY26 revenue forecast.
Conversion of the forecast 42 new customers or subscriptions, particularly the two dedicated-rack and 14 enterprise subscriptions.
Growth in contracted ARR and total contract value excluding SambaNova.
Disclosure of paying-customer numbers separately from trial users.
Rising utilisation and gross margin across the first ten racks.
Installation of the second ten-rack tranche from November 2026 and service availability in early 2027.
Evidence that enterprise and government customers accept SCX's sovereignty and security architecture.
A credible path toward EBITDA and operating cash-flow break-even.
A hardware-refresh roadmap that keeps pace with SambaNova and competing inference architectures.

Samso Concluding Comments

The easy way to present SCX.ai is to talk about the growth of artificial intelligence, the demand for sovereign infrastructure and the power constraints facing conventional GPU data centres.

All three themes are relevant. None of them, on its own, justifies the IPO valuation.

At the offer price, investors are paying approximately A$75 million for an operational company that has ten racks installed and forecasts A$6.04 million of FY26 revenue. After deducting pro forma cash, the prospectus enterprise value is approximately A$46 million. That is equivalent to about 7.6 times forecast revenue and 8.6 times contracted ARR.

There is substance behind the story. This is not a company with a slide deck and a plan to buy hardware later. The first node is installed. The platform is operating. The founder has already taken an enterprise-software company through the ASX. The A$40 million offer is fully underwritten, and the next ten racks are contractually committed.

There is also a concentration that cannot be ignored.

SambaNova supports most of the unbilled contract value and almost two-thirds of forecast FY26 revenue while also supplying the technology on which the platform depends. The offtake agreement is a useful bridge to utilisation, but it should not be mistaken for a diversified enterprise customer base.

The most important number after listing will therefore not be total ARR in isolation. It will be ARR from independent customers, accompanied by evidence that those customers are moving from trials into recurring production workloads.

SCX's model could produce attractive operating leverage if the existing racks fill. Cost of sales is substantially tied to installed capacity, so higher token consumption across the same hardware should improve margins. The inverse is also true. Underutilised racks still consume capital, data-centre services, support costs and management attention.

This is not a conventional SaaS story. It is a capital-intensive infrastructure and services business with a software layer. Investors must assess hardware economics, partner risk and customer utilisation as seriously as they assess ARR growth.

For investors comfortable with early-stage technology, infrastructure execution and customer-concentration risk, SCX offers a rare listed entry into Australia's emerging sovereign AI inference market.

For more cautious investors, the sensible approach is to watch the first reporting periods: separate partner revenue from independent customer growth, measure utilisation, and see whether the forecast sales pipeline converts before the second node materially expands the cost base.

The AI market may grow rapidly. SCX still has to prove that it can capture that growth on attractive economics.

That is where the value has to be created.

The Samso Way – Seek the Research

Here at Samso, we pride ourselves on delivering content for investors that is independent and informed by over three decades of experience in the industry. Our content is well-researched and is only created if I see merit in discussing the company's story.

Our mission is simple: cut through the noise and spotlight what matters—genuine stories, grounded insights, and real opportunity.

Our content is well-researched and is only created if the team sees merit in discussing the company or concept. Investors can explore our three core platforms:

There may be numerous paths to success in investing, but the common thread among successful individuals is that they remain committed to making informed decisions. Equip yourself with the right knowledge and tools, and you will be well on your way to achieving your financial goals.

Most importantly, investors need to be absolutely diligent in understanding their own risk-reward tolerance and capabilities. Never bite off more than you can chew. As they say, Rome wasn’t built in a day, and the Great Wall stood because it took centuries to complete.

The Samso Philosophy:

Stay curious. Stay sharp. And remember—digging deeper always uncovers the real value.

In Life, there is no such thing as a Free Lunch.

Never bite off more than you can chew is my parting comment.

Happy Investing, and the only four-letter word you need to know is DYOR.

To support our independent work, please head over to our Support Page and give us a helping hand in any of the ways listed. This is a new initiative for the Samso Platform, and it was always the concept of Samso when we started this journey in 2018.

Disclaimer

The information or opinions provided herein do not constitute investment advice, an offer, or solicitation to subscribe for, purchase, or sell the investment product(s) mentioned herein. It does not take into consideration, nor have any regard to your specific investment objectives, financial situation, risk profile, tax position and particular, or unique needs and constraints.

Read full Disclaimer.

About Samso

Samso is a trusted platform that equips dedicated investors with up-to-date industry knowledge and insights from top CEOs and thought leaders. By staying informed on business advancements and market trends, investors can enhance their financial decisions through a combination of expert guidance and their own research.

Samso News | www.samso.com.au | An Investor Lens on ASX-Listed Companies