Karsten Wenzlaff, Advisor
August 26th, 2025
September 29, 2026 | NCFA Insight | Artificial Intelligence And Data, Capital Markets And Market Infrastructure, Venture Funding And Building

On September 28, 2026, Reuters revealed new details from Anthropic's confidential IPO filing, showing 2025 revenue rose to about US$4.6 billion while compute and infrastructure costs reached US$7.33 billion. Anthropic has committed at least US$518 billion to cloud, compute and other AI infrastructure over roughly a decade.
Anthropic confidentially submitted its draft Form S-1 to the U.S. Securities and Exchange Commission on June 1. The filing isn't public yet, so the financial details come from journalists who reviewed the prospectus rather than an SEC document investors can inspect themselves. Anthropic also hasn't set the number of shares, IPO price or offering size yet.
The filing gives public market investors something private AI companies rarely provide. Anthropic's revenue is growing at extraordinary speed, but so are the costs and commitments needed to keep Claude running and improving. The question isn't simply whether demand for AI keeps rising. It is whether that demand can grow fast enough to absorb infrastructure contracted years before the customers arrive.
Anthropic generated about US$400 million of revenue in 2024 and nearly US$4.6 billion in 2025, an increase of roughly 11.5 times in one year. Compute and infrastructure spending moved just as aggressively, rising to US$7.33 billion in 2025, about three times the prior year level. That one expense represented roughly 58% of Anthropic's US$12.65 billion of operating expenses and about 1.6 times annual revenue.
| 2025 metric | Amount | What it tells investors |
|---|---|---|
| Revenue | ~US$4.6B | About 11.5 times 2024 revenue |
| Compute and infrastructure | US$7.33B | About 159% of revenue and 58% of operating expenses |
| Operating loss | US$8.06B | Operating costs still exceeded the rapidly growing revenue base |
| Net loss | ~US$42B | Includes a large financing related accounting charge |
| Cash and short term investments | US$20.28B | Year end liquidity before later financing |
| Future infrastructure commitments | At least US$518B | Multi year cloud, compute and infrastructure contracts |
The US$42 billion net loss needs a bit of context. Roughly US$34 billion came from an accounting charge tied to financing instruments that can convert into Anthropic shares. That doesn't mean Anthropic spent US$42 billion in cash operating Claude. The operating number is US$8.06 billion loss generated after revenue and operating costs are compared.
Does that mean the economics are getting worse as Anthropic grows? The current 2026 numbers make that difficult to conclude. Preliminary second quarter revenue exceeded US$11.5 billion, up from US$4.73 billion in the first quarter, and Anthropic reported positive adjusted operating income for the quarter. Those aren't audited full year results, but they show how quickly the picture is changing.
Rapid revenue growth doesn't remove the infrastructure problem, and it makes securing enough compute more urgent.
Anthropic expects to spend at least US$518 billion over roughly a decade across six infrastructure partners. Reuters reports that about 80% of those commitments are noncancelable or require payment regardless of how much capacity Anthropic actually uses. The known agreements include at least US$111.1 billion with Google, US$110 billion with Amazon, US$31.4 billion with Microsoft and about US$161.2 billion of Broadcom related equipment lease obligations.
Is US$518 billion debt?
No. The figure combines different cloud, compute, equipment and infrastructure contracts spread across years. Some arrangements can also be cancelled. Anthropic's agreements with xAI, for example, could involve up to US$84.5 billion of Nvidia based capacity through 2029 but are largely cancelable with 90 days notice.
The less flexible contracts create the bigger financial issue. Anthropic is reserving enormous amounts of future capacity before it knows exactly how quickly model efficiency, compute prices, customer demand and competition will change. Its filing reportedly warns that access to compute could constrain growth before customer demand does. If Anthropic is right, securing capacity early protects revenue growth. If demand, pricing or technology changes faster than expected, fixed infrastructure commitments become much harder to absorb.
BGC's compute infrastructure markets are creating price discovery and risk management around compute and memory capacity. When capacity is scarce, expensive and contracted years ahead, compute becomes something Anthropic has to finance and manage like major infrastructure, not just pay for as a cloud service.
The supplier relationships are unusual too. Amazon and Google are major Anthropic investors as well as infrastructure providers, and their platforms help distribute Claude. Microsoft also supplies infrastructure while competing across AI. These relationships give Anthropic capital, compute and distribution, but they also make a small group of large technology companies important to several parts of the business at once.
Demand is also concentrated. Nearly one quarter of 2025 revenue came from two customers, while many larger customers are not locked into long term contracts. Anthropic therefore has substantial fixed commitments on the infrastructure side while parts of its customer base remain comparatively flexible. That mismatch deserves your attention.
Reports that Anthropic may seek a valuation above US$2 trillion doesn't make the company worth US$2 trillion today. The firms last reported private valuation was about US$965 billion in May, so US$2 trillion remains a target until shares are actually priced.
Anthropic could also reach public markets before its closest frontier AI rival. OpenAI CEO Sam Altman said in September that OpenAI's public market timeline now extends beyond 2026, pushing any listing to at least 2027. Anthropic could therefore become the first major frontier model company to give public investors a direct look at the economics of building and operating AI at this scale.
Preliminary second quarter 2026 revenue exceeded US$11.5 billion, while Anthropic's annualized revenue run rate reportedly passed US$65 billion by the end of July. Investors will have to decide how much of that acceleration can persist and how much compute, capital and fixed infrastructure Anthropic must secure to support it.
Reuters reports that roughly 80 of the prospectus's 261 main pages cover risks, including the possibility that more capable AI could resist shutdown, conceal information, facilitate cyber or biological misuse and cause catastrophic harm.
Revenue can multiply in a year while infrastructure contracts extend for a decade. The market now has to decide whether that growth can remain fast and profitable enough to justify committing hundreds of billions of dollars to compute before future demand is known.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |
September 25, 2026 | NCFA Fintech Intelligence Question | Artificial Intelligence And Data, Risk Compliance And Regtech, Payments Infrastructure And Money Movement

On September 15, 2026, TypeSafe AI introduced Jev, an AI model built to make structured decisions inside software rather than write free form answers for people. Instead of generating prose, it returns predefined choices with probabilities and confidence scores.
TypeSafe calls Jev a System One Model and argues that this approach is better suited to autonomous software. That's the company's thesis, not an established industry conclusion. TypeSafe produced the benchmarks but acknowledges limitations in its testing and says Jev can still make wrong decisions.
AI agents can already read brokerage accounts, prepare trades, open wealth accounts, hold digital money and make payments. Giving an agent access is one thing. Letting it decide what to do with someone's money carries a different level of responsibility.
Do autonomous financial agents need a different AI architecture?
As agents progress from answering questions to spending, trading and making decisions, an error can become an action before a person sees it. If one AI interprets the request, makes the judgment and initiates the transaction, several distinct financial controls can end up depending on the same model.
Some firms are already separating what an agent understands from what it is allowed to do and what ultimately gets executed. The unanswered part is whether financial judgment should also be separated from the general purpose language model.
A specialized decision model is one option. An LLM with constrained outputs, external permissions and conventional software making the final call is another. There isn't enough independent testing in regulated financial workflows to know which approach performs better.
NCFA has already tracked AI's growing compliance burden, agents entering governed financial workflows and consent and liability when AI makes payments. Those controls become more consequential once an agent can act on a financial decision rather than simply recommend one.
Click each item to expand
Form3's Model Context Protocol adaptor gives authorized AI agents access to payment information through a controlled interface. Initial access is read only, and the agent stays outside the payment flow.
An agent can investigate a payment or help someone understand what happened without gaining the authority to send money. Form3 has separated access from execution at the infrastructure level.
OneVest launched agentic account opening for Canadian wealth firms on September 23. Its agents coordinate onboarding across custodians, track progress and handle administrative work that normally moves between people and systems.
Advisors and operations teams remain in control, and the system preserves an audit trail. The launch doesn't indicate that the agent independently approves a client or makes the final suitability decision.
GoCardless processed an AI driven account to account payment on September 22 through work developed in the FCA AI Live Testing programme.
The agent completed the transaction inside an AI conversation, but its authority came from an existing Direct Debit mandate. It could handle more of the transaction without receiving an open ended right to move money.
Mastercard, Flybits and Rogers Bank announced a completed Canadian agentic commerce transaction on September 21.
A Rogers Red Mastercard holder gave the Flybits powered assistant instructions and a spending limit. The agent selected an eligible product and completed the purchase using Mastercard Agent Pay.
The AI chose and paid, but only inside authority the customer had already defined.
TypeSafe describes Jev as its first System One Model. A developer defines the possible output types in advance. Jev returns a decision with probabilities and a measure of confidence rather than generating an unrestricted text response.
A response can therefore be perfectly formatted and still be wrong, an important distinction when the output can trigger a financial action.
Questrade's MCP connection lets approved AI agents work with brokerage account data and prepare orders when trading permission is enabled.
The customer still reviews the security, direction, order type, quantity, price, estimated cost and commission before submitting the trade. The agent can do much of the work without getting the final say.
Circle Agent Stack gives AI agents wallets that can hold and move USDC without asking a person to approve every transaction.
Developers can limit spending, restrict services, use allowlists and put time limits around an agent's authority. The agent may decide that it wants to pay, while the wallet determines whether that payment is permitted.
Click each item to expand
The Monetary Authority of Singapore and industry participants published Safeguards for Agentic Finance at Runtime, or SAFR, on July 3.
SAFR examines controls that operate while an agent is active, including whether its actions comply with policy, whether a transaction can be stopped, what records are kept and whether controls work across different systems.
Those safeguards operate outside the model rather than depending on the agent to police itself.
As financial agents gain access to accounts and transactions, governance extends beyond whether the model gives a good answer. Firms also need approved data, permissioned tools, review points, records, escalation and someone accountable when things go wrong.
Once an agent can act, the choice of which AI makes a financial decision becomes part of that control structure.
An April IMF Note divides agentic payments into three parts: intent, authorization and settlement.
AI can interpret what someone wants and turn that request into instructions. A separate authorization layer checks mandates, permissions and regulatory requirements before the approved transaction reaches payment infrastructure for settlement.
The IMF doesn't argue that LLMs should be replaced. Its model separates probabilistic AI from some of the controls governing the transaction.
There isn't one financial agent architecture yet. Form3 stops at information. Questrade lets an agent prepare a trade. OneVest automates much of an account opening. Circle lets agents spend within wallet rules. GoCardless and Mastercard take agents into live payments, but inside authority established before the transaction.
None of those examples establishes a need for specialized decision models. An LLM with strong permissions and conventional software controls may be enough for many jobs.
A fraud decision can be uncertain even when the permitted actions are tightly controlled. The available actions may be tightly defined even when the right answer isn't obvious: approve it, reject it, ask for more information or send it to a person. A model designed to choose among those outcomes and report its confidence could have a role between language interpretation and a hard software rule.
A model can perform well across thousands of decisions and still get one customer's transaction badly wrong. In credit, fraud, sanctions, trading or client money, the consequences of that one decision can matter more than the average benchmark.
Financial firms already separate customer instructions, risk decisions, permissions, execution and records. As AI enters governed financial workflows, firms have to decide which of those jobs AI should perform and which should remain outside its discretion.
Adoption by regulated firms would test TypeSafe's benchmarks against real credit, fraud, trading and payment decisions. Independent testing could then compare specialized models with LLM based agents and conventional systems, especially where a wrong decision can cost someone money.
Regulators may influence the architecture without choosing the model. Requirements that separate interpretation from authorization, limit an agent's authority or send uncertain decisions to a person could favour some designs over others. Those boundaries are already part of the debate around AI payment consent and liability.
LLMs may keep improving while structured outputs, wallet controls and policy engines handle the hard boundaries. Specialized models could prove valuable for particular financial decisions without becoming standard infrastructure. Understanding a customer's instruction is one job. Deciding whether their money should move can be another.
Do you agree the evidence supports this answer?
Click Agree or Disagree. Your vote is recorded anonymously and aggregate results are tracked.

The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |
September 25, 2026

Entrepreneurs are trained to manage uncertainty. They build forecasts with incomplete data, make decisions under pressure and keep teams moving through shifting market conditions.
Divorce creates a different kind of uncertainty.
Legal questions become entangled with parenting, housing, cash flow, business ownership and emotionally charged communication. Meanwhile, the company does not pause. Employees still need decisions, investors expect updates, customers require support and the next financing milestone remains on the calendar.
For founders, executives and technology professionals, divorce is not only a private family matter. It can become a leadership, financial and operational risk event. Managing it well requires more than working longer hours or applying the same intensity used to solve a product problem.
It requires the right advisory team, disciplined communication and a system for making consequential decisions without allowing personal conflict to consume the business.
A salaried employee may have income, savings and retirement accounts to consider. A founder’s financial life can include far more moving parts:
A company may appear valuable on a pitch deck without producing much personal liquidity. Conversely, an early-stage equity position that seems speculative today could become valuable after a financing, acquisition or public offering.
How these assets are treated depends on the facts and the applicable provincial or territorial law. A founder should not assume that shares registered in one person’s name, assets held digitally or property accumulated through a corporation are automatically excluded from the divorce process.
Early advice from qualified legal, tax and valuation professionals is essential.
Divorce involves several kinds of problems, and no single professional is meant to solve all of them.
A family lawyer provides legal advice and represents the client’s interests. An accountant or business valuator may analyze corporate income, ownership and value. A tax professional can identify the consequences of proposed transfers or settlement options. A therapist supports mental health and deeper emotional healing.
A divorce coach fills a different role. Coaching can help a client organize priorities, prepare for professional meetings, improve difficult communication and remain focused on practical next steps.
Busy professionals exploring this type of support can learn more from Heidi Dinning, a certified divorce coach and family communications expert who works with clients on divorce strategy, high-conflict communication, co-parenting and preparation for mediation or legal proceedings.
A coach does not replace a lawyer, accountant, valuator or therapist. The value comes from helping the client use each professional more effectively.
Founders rarely begin due diligence by searching through random email threads five minutes before a meeting. The same discipline should apply to divorce preparation.
Create a secure inventory of relevant personal and business records. Depending on professional advice, that may include:
This is an organizational step, not an invitation to copy confidential corporate information without authorization. Founders should ask legal counsel how to preserve relevant records while respecting company obligations, employee privacy and securities requirements.
Information should be collected transparently and lawfully. Concealing, deleting or transferring assets can create serious legal and credibility problems.
Divorce generates messages that feel urgent because they are emotionally uncomfortable. Not every message requires an immediate reply.
Entrepreneurs who are accustomed to rapid decision-making may respond too quickly, especially when a text or email feels unfair. That reaction can intensify conflict and create a written record that is difficult to explain later.
Before responding, ask:
A divorce coach can help a client review communication patterns, identify triggers and draft a clear response without turning every disagreement into a legal escalation.
This creates a valuable pause between reaction and action.
Founders make high-stakes decisions every day, but divorce can reduce the mental capacity available for those decisions. Sleep disruption, fear and repeated conflict make it harder to assess trade-offs objectively.
A useful approach is to divide decisions into three categories.
These can be adjusted with limited consequences, such as changing a weekly scheduling system or choosing where routine information will be shared.
These involve meaningful financial or parenting effects and should be considered carefully with the appropriate adviser.
Selling shares, transferring property, accepting a settlement or making public statements about the separation may create lasting consequences. These decisions should not be made simply to end an uncomfortable conversation.
The speed valued in a startup is not always an advantage in a divorce. Good leadership sometimes means slowing the process down long enough to obtain reliable information.
Personal disruption becomes a company problem when too many operational decisions depend on one person.
A founder navigating divorce should review what would happen if they were temporarily unavailable. This does not require sharing private details with the entire team. It requires ensuring that the company is not dependent on the founder answering every question in real time.
Consider:
If the divorce may affect ownership, governance, financing or disclosure obligations, the appropriate corporate and family-law advisers should be involved. The board may also need information when the issue creates a material risk, but disclosure should be deliberate and limited to what is professionally required.
Technology professionals often have a complicated digital footprint. Personal and corporate data may be spread across cloud storage, password managers, shared subscriptions, smart-home systems, cryptocurrency wallets and devices used for both work and family communication.
During separation, review digital access carefully without locking a spouse out of jointly owned property or interfering with information that must be preserved.
Useful steps may include:
Company IT staff should not be drawn into a personal dispute. If corporate security is potentially affected, involve the appropriate security or legal leader through a controlled process.
Lawyers, accountants and valuators are valuable because of their specialized expertise. Meetings are more productive when the client arrives with organized documents, clear questions and a defined objective.
Before each professional meeting, prepare a short brief:
A divorce coach can help build this agenda and organize the emotional noise surrounding the issue. This allows legal and financial professionals to concentrate on the work only they are qualified to perform.
Entrepreneurs are used to developing solutions quickly. Children experiencing a family separation may not need a solution in the moment. They may need time, reassurance and permission to have complicated feelings.
A parenting plan is not simply another operating agreement. It affects a child’s routines, relationships and sense of security.
Avoid treating time with children as a metric to be maximized or a negotiation to be won. Where it is safe and appropriate, keep children out of adult conflict, avoid using them as messengers and do not ask them to report on the other household.
Co-parenting coaching can help parents prepare for difficult conversations, establish workable communication practices and keep decisions centred on their children’s needs.
Legal complexity is not the only factor that makes divorce expensive. Repeated conflict can generate unnecessary correspondence, emergency meetings and negotiations over issues that might otherwise have remained manageable.
Watch for patterns such as:
A structured communication process can reduce confusion. Keep messages brief, address one subject at a time and separate factual requests from emotional commentary. When a response is not urgent, draft it and review it later.
The first days of separation can feel chaotic. A simple sequence can restore a degree of control:
A successful divorce strategy is not about eliminating emotion. That is unrealistic. It is about preventing emotion from driving every legal, financial and parenting decision.
For a busy founder or technology leader, the right support structure protects more than personal well-being. It preserves attention, improves the use of professional advice and reduces the likelihood that a private crisis will become an unmanaged business risk.
Entrepreneurs already understand the value of expert guidance in areas outside their specialization. Divorce should be approached with the same discipline. Build the team, organize the information, protect the company and create enough space to make decisions based on facts rather than fear.
This article provides general information only and does not constitute legal, financial, tax, investment or mental-health advice. Divorce and property laws vary by jurisdiction. Obtain advice from qualified professionals who understand your circumstances.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |