AI Fugazi: When Real Technology Rests on Synthetic Financial Plumbing

AI Fugazi: When Real Technology Rests on Synthetic Financial Plumbing

By Dr Danie Adendorff

A legally cautious analysis of whether the financial architecture around frontier AI is beginning to reproduce the opacity, circularity, maturity mismatch and valuation discretion that have preceded earlier financial dislocations.

Opening argument: the technology is real; the plumbing is the question

The most responsible way to write about the present AI boom is neither to sneer at it nor to submit to it. Artificial intelligence is real. The demand for compute is real. Nvidia hardware is real. xAI’s data-centre ambitions are real. Apollo-linked financing of AI hardware is real. Athene’s role inside Apollo’s broader retirement-services and private-credit ecosystem is real. The question is not whether the objects exist. The question is whether the financial representation of the boom is becoming more synthetic, more circular and less transparent than the underlying economics justify.

That is where the phrase “AI Fugazi” earns its analytical purchase. “Fugazi” is used here in its colloquial sense: something that may appear solid, clean or authentic on the surface while concealing an artificial or questionable underlying construction. It is not used as a legal allegation of fraud. The argument is narrower and more defensible: parts of the frontier-AI capital stack increasingly exhibit features historically associated with fragility — vendor-linked demand, lease-mediated leverage, special-purpose vehicles, private-credit intermediation, model-valued assets, offshore or insurer-adjacent funding channels, and opacity over the final bearer of economic risk.

The title does not depend on attributing the word “fugazi” to Michael Burry or to any other market participant as a verified quotation. It is used as the author’s analytical label for a narrower condition: the possible divergence between real technological capability and the financial representation built around that capability.

The prudent conclusion is therefore not “AI is fake”. It is that a real technology can still sit on a financial structure whose stability depends on continued confidence, continued hardware scarcity, continued revenue growth, continued refinancing capacity, and continued willingness by private-credit and insurance-linked channels to absorb illiquid exposures at scale. That is not a prophecy of collapse. It is a diagnosis of a risk condition.

The verified transaction pattern

The strongest factual record concerns xAI-linked financing for Nvidia compute infrastructure. In July 2025, Reuters, citing the Wall Street Journal, reported that xAI was working with Valor Equity Partners to raise up to $12 billion in debt to buy Nvidia chips that would be leased to xAI for a large data-centre build-out. In October 2025, Reuters, citing Bloomberg, reported that xAI was nearing a financing package of up to $20 billion, structured around roughly $7.5 billion of equity, up to $12.5 billion of debt, and Nvidia participation of up to about $2 billion, with the capital tied to GPU purchases through a special-purpose vehicle. In February 2026, Reuters reported that Apollo was close to a roughly $3.4 billion loan to an investment vehicle that would buy Nvidia chips and lease them to xAI. The same report described this as Apollo’s second major investment in a chip-leasing venture for xAI, following a $3.5 billion loan in November 2025.

The important point is not that leasing is inherently improper. It is that the AI infrastructure build-out is not simply being funded as conventional balance-sheet capital expenditure by a profitable industrial firm buying equipment for its own long-term use. A significant part of the model appears to involve investor vehicles, debt finance, depreciating hardware, lease contracts, sponsor economics, vendor participation and private-credit capital. This is a different kind of financial architecture, and it deserves a different kind of scrutiny.

Why rational infrastructure finance can still become reflexive

A fair analysis must begin with the strongest counterargument. Lease-backed infrastructure finance is not new. Data-centre operators, telecom-tower companies, aircraft lessors, equipment-finance firms and real-estate investment vehicles have long used structures in which capital providers own or finance assets that operating companies rent over time. Such structures can be economically rational. They allow fast-growing firms to access productive assets without exhausting cash reserves. They create identifiable collateral for lenders. They can transfer asset ownership to investors better placed to absorb duration and residual-value risk. They can also widen the capital base for infrastructure that would otherwise be too expensive for a single balance sheet.

The term “reflexive” is used here in the broad financial sense associated with George Soros’s theory of reflexivity: market participants do not merely observe fundamentals; their beliefs, financing choices and valuation practices can help change the fundamentals they claim to measure. In an AI capital cycle, optimistic expectations about compute demand can mobilise financing, that financing can generate visible orders, and those orders can then appear to validate the original optimism.

The AI case is different because the financed asset is not a stable building, a tower or a long-life utility asset. It is high-cost compute hardware exposed to rapid technological obsolescence, uncertain useful life, intense demand cyclicality and uncertain end-user monetisation. The customer base is concentrated, the revenue model remains contested, the hardware supply cycle is strategically sensitive, and the valuation narrative is unusually dependent on the belief that frontier AI will keep absorbing more compute at economically profitable rates.

This is where the new empirical context matters. Financial Times reporting indicates that major technology groups have shifted more than $120 billion of AI-related data-centre financing off balance sheets through special-purpose vehicles and Wall Street-backed structures. That figure should not be read as proof of abuse. It should be read as a scale indicator: a financing form that may be rational at transaction level has become sufficiently large to require system-level scrutiny.

Quinn Emanuel’s March 2026 litigation-risk analysis adds a useful legal lens. It identifies financing mechanics including direct loans, SPV structures, securitisations and GPU-collateralised facilities, and it flags risks around defaults, insolvency cascades, off-balance-sheet opacity, credit-ratings disputes, GPU-collateral valuation, take-or-pay obligations and multi-party contractual conflict. That is not evidence that any specific xAI, Nvidia, Apollo or Athene transaction is unlawful. It is evidence that lawyers now see AI infrastructure finance as a zone where complex capital structures may generate future disputes if demand, collateral value or refinancing assumptions weaken.

A lease structure that is ordinary in form can therefore become extraordinary in risk when the collateral is technologically perishable, the cash-flow model is still maturing, and the structure is embedded in a market where off-balance-sheet financing is expanding rapidly. This is the central distinction: the legal wrapper may be familiar, but the economic conditions around frontier AI are unusually volatile.

Vendor-financed demand and the interpretation problem

The Nvidia element sharpens the issue. Reuters reported, citing Bloomberg, that Nvidia could contribute up to about $2 billion to xAI financing tied to GPU acquisition. Reuters separately described Nvidia as an anchor investor in the Apollo-backed compute vehicle. Even if each transaction is commercially defensible, the optics are analytically material: the supplier is not merely selling the picks and shovels; it may also be helping capitalise the gold rush that buys them.

That does not prove impropriety. It does, however, complicate interpretation of reported demand. When a vendor participates financially in a customer’s ability to purchase the vendor’s own product, revenue can remain real while the market signal becomes harder to read. The question becomes whether demand is being revealed by independent customer economics or partly reinforced by a circular capital loop. Demand justifies financing; financing enables orders; orders validate the supplier narrative; the supplier narrative attracts more capital; more capital enables further demand. That loop can be rational during expansion. It can also become fragile if residual values, useful-life assumptions, refinancing conditions or AI monetisation disappoint.

Disclosed does not mean transparent

The Athene/Apollo dimension is consequential and must be handled carefully. Athene’s own materials present a formal defence: Apollo and Athene maintain separate capital structures; Athene identifies regulated entities in Iowa, Bermuda and Delaware; its corporate-structure materials state that Bermuda entities help raise third-party capital efficiently; and those materials also state that Athene has no regulated insurance entity in the Cayman Islands or in jurisdictions not deemed equivalent by US regulators. These facts matter and should not be suppressed.

At the same time, Athene’s filings and external reporting show why analysts focus on opacity, valuation discretion and affiliated origination. Athene’s Q3 2025 filing reported about $429.9 billion of total assets, with large fair-value exposures including roughly $130.8 billion of Level 3 assets and substantial related-party positions. On a simple total-assets denominator, $130.8 billion against $429.9 billion is approximately 30.4%, not 36%. The separate Financial Times-reported 36% figure should therefore not be read as a direct reconciliation to those two filing figures; it may reflect a different denominator, such as invested assets or another portfolio measure rather than total assets. Because the specific FT denominator was not independently verified from the reviewed public record, the 36% figure is retained only as a B2-B3 source claim and not as an independently verified primary-filed ratio.

The public record reviewed for this article does not establish that Athene policyholder-related capital directly funded the specific xAI-linked chip-leasing vehicles. That inferential gap is important. The defensible claim is not that Athene bought those securities. The defensible claim is that the same broader Apollo-linked and insurer-adjacent private-credit ecosystem sits close to the capital channels now being used to finance frontier-AI infrastructure. That proximity is enough to justify scrutiny; it is not enough to justify a direct actor-specific allegation.

This distinction is the heart of the matter. A system can be formally disclosed and still be difficult for outsiders to understand. “Disclosed” means that the relevant legal and regulatory documents may exist. “Transparent” means that outside creditors, policyholders, regulators, analysts and investors can see who bears downside risk, how assets are valued, how correlated exposures are, how refinancing will work under stress, and whether legal form matches economic substance. Those are not the same standard.

Model-valued assets and the valuation problem

The term “model-valued assets” requires precision. In this context, it refers primarily to assets whose reported value depends less on observable market prices and more on internal models, assumptions, discounted cash-flow estimates, comparable transactions, broker inputs or managerial judgement. Level 3 fair-value assets are the clearest example: they are not necessarily bad assets, but they are harder for outsiders to price, harder to liquidate under stress, and more vulnerable to optimism in assumptions about default risk, discount rates, collateral quality and future cash flows.

AI finance adds a further valuation difficulty. Some financed assets are tangible — GPUs, servers, data-centre equipment, power infrastructure and lease contracts. But the repayment narrative is tied to less tangible assumptions: model capability, user adoption, enterprise willingness to pay, inference economics, competitive position, hardware scarcity and future refinancing capacity. The risk is not that accountants simply invent numbers. The risk is that a chain of plausible assumptions can produce a valuation architecture that looks disciplined in expansion but becomes difficult to defend when technology cycles, customer economics or credit conditions turn. In frontier AI, the line between collateral value, strategic value and narrative value can blur quickly.

The depreciation mechanism is particularly important. If a lease vehicle assumes that GPU fleets will retain useful economic value over a period that later proves too optimistic, the first effect may be a residual-value problem rather than an immediate default. But residual-value pressure can then travel through the structure: collateral coverage weakens, refinancing terms tighten, lenders dispute marks, lessees resist renewal pricing, and sponsors may be forced either to inject support or accept impairment. In that sequence, technological obsolescence becomes a credit event by indirect route.

Why historical analogues matter — and where they do not

No analogy should be treated as identity. The present AI build-out is not Enron, Lehman, AIG, Greensill, Archegos or the subprime mortgage chain in literal form. The hardware is real. The contracts are real. The companies are real. The lenders are sophisticated. The public record does not prove deception. The point of historical comparison is narrower: previous financial blow-ups often began not with imaginary assets, but with structures that converted complexity into apparent safety while pushing economic risk into places outsiders could not see.

Enron is relevant not because today’s AI vehicles are equivalent to Enron’s abusive special-purpose entities. They are not. Enron’s structures were embedded in accounting deception and corporate fraud. The more limited lesson is that when economics are split across sponsors, vehicles, counterparties and affiliated channels, outside observers can misread where the downside ultimately sits. The warning is about opacity and incentive alignment, not a claim that AI leasing vehicles are operationally fictitious.

Lehman’s Repo 105 is relevant for a related reason. The problem was not that repurchase markets did not exist or that repo was inherently illegitimate. The problem was that legal and accounting form could be used to shape perceptions of leverage. The AI analogue is not an accounting match. It is the danger of allowing a transaction’s form — lease, SPV, vendor investment, private-credit vehicle — to overpower analysis of its economic substance.

AIG Financial Products demonstrates how seemingly remote, well-rated and modelled exposures can become a liquidity crisis once collateral calls, downgrades and confidence effects interact. Hardware leases and credit derivatives are different instruments, but both can be destabilised when correlation is underestimated and when liquidity assumptions depend on benign conditions continuing.

Greensill offers a more immediate lesson about the seduction of future-looking receivables. Supply-chain finance was not inherently illegitimate. The fragility emerged when the model stretched toward prospective receivables, insurance reliance, concentration risk and funding confidence. The AI-finance parallel lies in the danger of converting a future-growth story into present-day yield product before cash-flow quality has been fully tested.

Archegos shows how risk can become systemically relevant even when no single counterparty sees the entire exposure map. Total-return swaps concealed leverage and concentration across prime brokers until margin pressure forced liquidation. The AI-finance concern is similar in structure, not instrument: sponsors, vendors, SPVs, lenders, insurers and end-investors may each see a manageable slice while nobody outside the core arrangement has a full map of concentration, maturity, collateral and refinancing risk.

The private-credit and insurance nexus

The most revealing current comparator is therefore not the 2008 mortgage itself but the private-credit-and-insurance nexus now under supervisory scrutiny. Reuters reported that Moody’s found US life insurers moved nearly $800 billion in reserves offshore between 2019 and 2024, largely to Bermuda and the Cayman Islands, amid a shift toward private credit. The Bank of England’s Prudential Regulation Authority has also moved to tighten the prudential treatment of funded reinsurance through CP8/26, proposing changes to how such arrangements are valued on Solvency UK balance sheets.

These regulatory developments matter because they show that supervisors are not merely concerned with isolated transactions. They are concerned with patterns: long-dated liabilities, illiquid assets, offshore reinsurance, affiliated origination, private-credit growth, valuation discretion and the possibility that headline solvency measures may not fully capture stress dynamics.

The second empirical upgrade concerns private-credit stress. According to With Intelligence’s proprietary 2026 private-credit outlook, headline private-credit default rates have remained below 2% for several years, but once selective defaults and liability-management exercises are included, the “true” default rate approaches 5%. This is treated here as a B2-B3 source claim, not as a primary regulatory statistic. Reuters has separately reported that unrealised losses at US private-credit lenders deepened in the first quarter of 2026 and that payment-in-kind income remained elevated. These are not direct measurements of AI-infrastructure credit impairment. They are warning indicators from the funding environment into which AI infrastructure finance is expanding.

AI infrastructure finance enters this environment with an awkward profile: it is capital-intensive, fast-moving, narrative-rich, concentrated, technology-dependent and hungry for long-duration funding. A resilient private-credit market can absorb some of that demand. A late-cycle private-credit market facing higher adjusted default indicators, PIK reliance and valuation pressure may demand tighter terms or become less willing to warehouse illiquid exposure. That is the relevance of the 5% adjusted-default signal: not that it proves AI finance is failing, but that it raises the cost of assuming private-credit capital will remain indefinitely abundant and forgiving.

Lawful does not mean clean

The ethical issue is not limited to legality. A structure can be lawful, disclosed, professionally arranged and still be problematic if it migrates risk to parties who cannot realistically see, price or contest that risk. Policyholders, pension-linked investors, credit-fund investors and public-market shareholders may not experience the transaction as a frontier-AI opportunity. They may experience it as a claim on retirement capital, insurer reserves, fund returns or market confidence.

This is why the language of “AI Fugazi” is useful only if disciplined. It should not be used to imply proven fraud. It should be used to name a condition in which technological reality and financial representation begin to diverge. The technology may improve. The models may become more useful. The data centres may be necessary. Nvidia may continue to be a dominant supplier. None of that proves that every financing structure built around the boom is resilient. A real technology can still support a synthetic confidence machine.

What should be monitored

The article’s conclusion should not be a prediction of collapse. It should be a monitoring framework. The following indicators would matter most over the next 12 to 24 months:

Off-balance-sheet displacement.

Track whether AI data-centre and hardware spending continues moving into unlisted SPVs, private-credit vehicles or other structures that reduce visibility in operating-company accounts. The FT-reported $120 billion figure should be treated as the current public scale marker, not as a final ceiling.

Private-credit portfolio stress.

Track whether adjusted private-credit default indicators move decisively beyond the approximate 5% level identified by With Intelligence once selective defaults and liability-management exercises are included, and whether Reuters-reported unrealised losses and PIK reliance continue to rise.

GPU depreciation discipline.

Track whether firms extend useful-life assumptions beyond realistic obsolescence boundaries in ways that smooth earnings, protect asset marks or delay recognition of residual-value impairment. In operational terms, the issue is whether accounting lives, lease tenors and collateral marks remain credible when a new GPU generation materially changes performance-per-watt, inference economics or resale value.

Vendor-financed demand.

Track the proportion of reported hardware demand supported directly or indirectly by supplier equity, strategic investment, customer financing, reciprocal capital loops or take-or-pay arrangements.

Lease concentration and refinancing exposure.

Track whether a small number of AI tenants, sponsors, hardware suppliers or private-credit lenders become central to a large share of the same underlying compute-finance exposure.

Insurer-linked capital migration.

Track the volume of capital backing AI infrastructure leases or private-credit exposures that sits in offshore reinsurance, affiliated origination, insurer-adjacent credit funds or hard-to-model Level 3 portfolios.

Regulatory capital interventions and litigation signals.

Track the final shape of the PRA’s funded-reinsurance rules, any SEC or state-insurance scrutiny of off-balance-sheet AI infrastructure finance, and the emergence of disputes over GPU collateral values, SPV disclosure, credit ratings, take-or-pay contracts or refinancing failures.

None of these indicators alone would prove failure. Together, however, they would reveal whether the AI capital stack is becoming more transparent and resilient, or more recursive and fragile.

Conclusion: financial reflexivity under technological euphoria

The right conclusion is not that AI is fake. The right conclusion is that parts of the AI capital stack are beginning to look like a confidence system layered on top of a real technology. That distinction is essential. The technology may endure. The products may improve. The supplier may keep winning. But the financing superstructure around the build-out can still become unstable if it relies on recursive validation: investors fund hardware because demand looks unstoppable; demand looks unstoppable because investors fund hardware; reported orders validate the supplier story; the supplier story validates the investment thesis; and accounting, ratings and portfolio marks smooth the transition from present liquidity to future uncertainty.

A market does not need criminal fraud to become fragile. It needs only a widening gap between narrative value and cash-flow truth. If GPU useful lives shorten faster than assumed, if newer architectures devalue installed fleets, if frontier-model revenues fail to catch up with capital intensity, if private-credit stress rises, if regulatory pressure raises the cost of insurer-linked funding, or if off-balance-sheet SPVs face disputes over collateral, disclosure or refinancing, a structure that looks ingenious in expansion can look circular in contraction.

That is the warning. AI may be real while parts of its financial architecture become synthetic. The proper name for that condition is financial reflexivity under technological euphoria. If one wants a sharper phrase, “AI Fugazi” will do.

Source notes and evidential guardrails

This article distinguishes between verified fact, supported inference and unverified allegation. Verified fact refers to company materials, SEC filings, official regulatory documents or high-quality reporting directly describing a transaction or regulatory development. Supported inference refers to structural conclusions drawn from multiple verified facts, such as the resemblance between AI lease-financing loops and earlier episodes of financial reflexivity. Unverified allegation refers to claims for which no public primary-source support was located, including direct Athene ownership of the specific xAI vehicles and the exact “fugazi” wording attributed to Michael Burry.

The article’s strongest verified claims are that xAI-linked Nvidia infrastructure financing has used debt, SPV and lease structures; that Nvidia was reported as a financial participant in some structures; that Apollo-linked financing was reported in connection with xAI chip-leasing vehicles; that Athene is central to Apollo’s retirement-services ecosystem and has large Level 3 and related-party exposures; that more than $120 billion of AI data-centre financing has reportedly moved off balance sheet; and that regulators and market analysts are scrutinising funded reinsurance, private credit and insurance-linked capital. The article does not assert that Athene directly funded the specific xAI-linked vehicles. It also does not assert that Michael Burry used the word “fugazi” in a verified public primary-source statement about this transaction set.

Source confidence guide: A1 refers to primary official documents or filings; A2 refers to high-quality factual reporting from established news agencies; B2-B3 refers to credible but secondary, paywalled, proprietary or methodology-limited reporting that should be attributed explicitly and not treated as independently verified primary evidence.

Selected references

1. Reuters. “Musk’s xAI to raise up to $12 billion in debt for AI expansion, WSJ reports.” 22 July 2025. https://www.reuters.com/business/musks-xai-raise-up-12-billion-debt-ai-expansion-wsj-reports-2025-07-22/

2. Reuters. “Musk’s xAI nears $20 billion capital raise tied to Nvidia chips, Bloomberg News reports.” 7 October 2025. https://www.reuters.com/business/musks-xai-nears-20-billion-capital-raise-tied-nvidia-chips-bloomberg-news-2025-10-07/

3. Reuters. “Apollo, xAI near $3.4 billion deal to fund AI chips, The Information reports.” 9 February 2026. https://www.reuters.com/business/apollo-xai-near-34-billion-deal-fund-ai-chips-information-reports-2026-02-09/

4. Financial Times. “Tech groups shift $120bn of AI data centre debt off balance sheets.” 2026. https://www.ft.com/content/0ae9d6cd-6b94-4e22-a559-f047734bef83

5. Quinn Emanuel Urquhart & Sullivan. “Emerging Litigation Risks in Financing AI Data Centers Boom.” March 2026. https://www.quinnemanuel.com/media/4dzkfccz/client-alert-ai-data-center-financing-and-litigation-risks.pdf

6. Athene Holding Ltd. Investor relations and SEC filings, including 2024 corporate-structure materials and Q3 2025 Form 10-Q. https://ir.athene.com/

7. Athene Holding Ltd. Form 8-K exhibit, “Overview of Athene’s Corporate Structure.” Filed 17 June 2024. https://www.sec.gov/Archives/edgar/data/1527469/000152746924000049/ahl-20240617.htm

8. Athene Holding Ltd. Form 10-Q for the quarter ended 30 September 2025. Filed 10 November 2025. https://www.sec.gov/Archives/edgar/data/1527469/000152746925000079/ahl-20250930.htm

9. Financial Times. “How insurance became the lifeblood of private credit.” 23 February 2026. B2-B3 treatment for the reported 36% Level 3 figure. https://www.ft.com/content/b6be87a9-0abf-4950-9060-2159aa547f3d

10. Reuters. “US life insurers shifted $800 billion offshore from 2019 to 2024, Moody’s says.” 2 June 2025. https://www.reuters.com/world/americas/us-life-insurers-shifted-800-billion-offshore-2019-2024-moodys-says-2025-06-02/

11. Bank of England, Prudential Regulation Authority. CP8/26, “Funded reinsurance.” 29 April 2026. https://www.bankofengland.co.uk/prudential-regulation/publication/2026/april/funded-reinsurance-consultation-paper

12. With Intelligence. “Private Credit Outlook 2026: The Market Faces its First Big Test.” May 2026. https://www.withintelligence.com/insights/private-credit-outlook-2026/

13. Reuters. “Unrealised losses at US private credit lenders deepen.” 29 May 2026. https://www.reuters.com/business/finance/unrealised-losses-us-private-credit-lenders-deepen-2026-05-29/

14. U.S. Securities and Exchange Commission. “Report and Recommendations Pursuant to Section 401(c) of the Sarbanes-Oxley Act of 2002 On Arrangements with Off-Balance Sheet Implications, Special Purpose Entities, and Transparency of Filings by Issuers.” 2005. https://www.sec.gov/news/studies/soxoffbalancerpt.pdf

15. Valukas, A. R. “Report of Anton R. Valukas, Examiner, Lehman Brothers Holdings Inc.” 2010.

16. Financial Crisis Inquiry Commission. “The Financial Crisis Inquiry Report.” 2011. https://www.govinfo.gov/content/pkg/GPO-FCIC/pdf/GPO-FCIC.pdf

17. House of Commons Treasury Committee. “Lessons from Greensill Capital.” 20 July 2021. https://committees.parliament.uk/publications/6733/documents/71882/default/

18. Reuters. “Archegos’ Bill Hwang sentenced to 18 years in prison for massive US fraud.” 20 November 2024. https://www.reuters.com/legal/archegos-bill-hwang-be-sentenced-massive-us-fraud-2024-11-20/

19. Reuters. “Michael Burry of Big Short fame deregisters Scion Asset Management.” 13 November 2025. https://www.reuters.com/sustainability/sustainable-finance-reporting/michael-burry-big-short-fame-deregisters-scion-asset-management-2025-11-13/

20. Reuters. “AI bubble? Opinions divided on tech’s trillion dollar question.” 16 October 2025. https://www.reuters.com/business/finance/opinions-split-over-ai-bubble-after-billions-invested-2025-10-16/

Author workflow disclosure

This article was produced through an AI-assisted but human-directed workflow. AI support was used for accessibility assistance, article structuring, language refinement, source-discovery prompts, revision planning, and conversion of editorial comments into amendments. Dr Danie Adendorff retained responsibility for the argument, accepted or rejected changes, checked the logic of claims, assessed source credibility, and remains accountable for the final text. AI-generated material was not treated as empirical evidence, and synthetic or illustrative examples were not presented as observed data.

Image disclosure:

The image accompanying this article was generated using artificial intelligence and is provided for illustrative purposes only. It does not depict an actual event, location, or documented scene.

contact@rumbls.com

Terms and Conditions

Privacy Policy

Cookie Policy

© 2026 Dr Danie Adendorff. All rights reserved.

Disclaimer