The $335m Signature That Never Became Cash

Founders often treat a signed foreign investment agreement as the finish line, then discover that screening rules, bank checks, corporate procedures and payment documentation still control the wire. Cases from Cerebras to MTN Nigeria show that the last uncleared gate, not the signature, determines when capital becomes usable cash.

Cerebras Systems, the California company building AI chips, had a signed preferred-stock agreement worth $335 million.

Its buyer was G42, an Abu Dhabi company. Under the May 2024 agreement, G42 was to buy roughly 22.9 million Cerebras shares. The first deadline was April 15, 2025.

That sounds like a completed financing.

It was not.

The companies voluntarily filed with the Committee on Foreign Investment in the United States, or CFIUS, the US body that reviews certain foreign investments for national-security concerns. They later changed the proposed shares from voting to non-voting, then requested withdrawal of the filing. CFIUS granted the withdrawal on March 27, 2025.

Still, the purchase had not completed by April 15.

Cerebras’ registration statement dated April 17, 2026 disclosed that G42 had been removed from the agreement in the third quarter of 2025 and that the related purchase option had been terminated. The $335 million did not become equity cash for Cerebras.

Cerebras had a signed $335 million agreement, yet the proposed G42 investment never became equity cash. Photo: Coolcaesar / Wikimedia Commons, CC BY 4.0.

At first glance, this is a story about chips and geopolitics. It is also a familiar fundraising error in unusually sharp relief: treating a signature as if it were money.

Ask the question that matters. If your overseas investor signed this afternoon, could the cash arrive when payroll, an acquisition deadline or your runway requires it?

For many cross-border raises, the honest answer is: not yet.

The signature myth

The usual story goes like this: find the investor, agree a valuation, sign the papers, then let lawyers complete the administration.

That gets the order wrong.

A foreign commitment is a chain of permissions. The investor may need authority to send capital. The recipient company must be able to issue the agreed shares. A regulator may review the ownership. Sending and receiving banks must accept the people, source, purpose and documents behind the payment.

Any one can stop the deal.

A Delaware company can be useful because its board can authorise a share issue. A UK private company can report an allotment on an SH01 within one month. Those are company-law steps.

They do not erase CFIUS scrutiny, UK national-security screening, foreign capital-transfer restrictions, sanctions checks or bank compliance.

This is where most people stop looking. They ask whether the company can issue shares. They do not ask whether every institution required to move the money has accepted the transaction.

GI Network's view: A cross-border commitment should enter a cash forecast only after its investor, issuer, regulator and bank gates have an evidenced route to clearance, with dates and named owners.

Non-voting did not mean non-sensitive

Cerebras challenges a comforting assumption: if foreign ownership is the concern, remove the vote and the problem disappears.

It may reduce governance rights. It does not automatically remove national-security risk. Nor does it create an executable closing.

The US Treasury says a CFIUS process can include a 45-day review, a further 45-day investigation and 15 days for presidential action. Investigated cases in 2023 averaged 85.8 calendar days.

Cerebras and G42 changed the security and withdrew their filing. Yet the transaction did not close before its deadline. The missing ingredient was not a cleverer share right. It was a fully executable alternative: funded escrow, a replacement investor or a specified non-equity arrangement that could actually settle.

Imagine hiring against expected cash, telling suppliers the round is signed, then discovering that the structure intended to unlock the deal has been withdrawn. Without a funded back-up, the cash forecast was fiction when it mattered.

The fund that could not buy its prize

The next gate can sit further down the ownership chain.

Leo Investments, a Hong Kong-based company owned by a listed Chinese group, committed $50 million to a Tomales Bay Capital fund formed to buy SpaceX shares. The fund documents were signed on November 15, 2021.

Then SpaceX, the US company with government-contract exposure, learned of the Chinese-backed investor.

It objected, raising concerns including possible CFIUS review and consequences for government contracts. Within roughly a week, Leo was removed and its $50 million was returned.

The dispute reached Delaware’s Chancery Court, which awarded nominal damages on the substantive dispute. Related appeals continued in 2026.

SpaceX’s objections showed that a fund investor may still be unable to reach the underlying asset. Photo: Bruno Sanchez-Andrade Nuño from Washington, DC, USA / Wikimedia Commons, CC BY 2.0.

The operating lesson is brutal in its simplicity. Leo had entered the fund. Yet the fund could not give it access to the underlying asset.

A special-purpose vehicle, or SPV, is a separate company or fund built for one investment. It can organise ownership. It cannot override restrictions imposed by the company inside it.

Put Cerebras and Leo side by side and a pattern appears that neither case says outright: ownership risk travels through the chain. It does not politely stop at the entity nearest the investor.

That is why a founder taking money through a fund, feeder vehicle or nominee needs to know the real investor behind it, the target company’s restrictions and every consent needed further down the chain. Proof of funds can prove almost nothing if the capital cannot enter the asset you are actually financing.

India’s clock starts when the money arrives

The hidden gate is not always sensitive US technology. Sometimes it is the routine machinery of receiving foreign money.

In India, the authorised-dealer bank is part of the foreign-investment process, not merely the courier for a wire. Sector eligibility, pricing rules, government-route approval where required, receipt of payment, share allotment and reporting through that banking channel must line up.

After consideration is received, shares generally must be issued within 60 days, or the money refunded within 15 days. That turns sequencing into a legal issue, not a diary-management preference.

Great Wall Motor, the Chinese carmaker, learned the larger version of this lesson in its planned purchase of General Motors India’s Talegaon plant. On January 17, 2020, Great Wall and GM signed a binding term sheet and expected completion in the second half of that year, subject to approvals.

The proposed plant payment was reported at up to $300 million within a wider $1 billion investment plan. Those figures suggest why the transaction mattered to Great Wall’s proposed Indian entry. They do not, on their own, prove that the commercial case was strong.

Then India tightened screening of investments linked to neighbouring countries. Repeated extensions followed. The agreement expired on June 30, 2022, 29 months after signing, because approvals had not been obtained. Great Wall abandoned both the acquisition and its planned manufacturing entry.

Great Wall’s planned entry into India stalled as approval rules changed after the binding term sheet. Photo: RAJESH KUMAR VERMA / Pexels, Pexels licence (free commercial use).

What surprised us was not that approvals took time. It was that the rulebook moved while the buyer waited.

Great Wall’s experience is not evidence that every India-linked transaction will stall. It shows that approval risk is not fixed on signing day. A business plan built around one unapproved asset needs more runway than the headline timetable suggests, and preferably another route into the market.

There is a useful contrast in South Africa. Mauritius-registered ECP Africa funds sought 95.78% of Burger King South Africa and all of its meat plant. The Competition Commission received notification on March 4, 2021 and prohibited the transaction on June 1 because it would adversely affect ownership by historically disadvantaged persons.

The parties negotiated with the Commission and government. The Competition Tribunal approved the acquisition with conditions on September 17.

That is not an instant close. But it is a named example of screening that was manageable because the obstacle became concrete, the parties reworked the transaction around the statutory public-interest issue, and the approval arrived within months rather than drifting into years. The difference was not that regulation vanished. It was that the deal found an acceptable shape.

Approval can rewrite the economics

Some gates do not kill a deal. They alter what the money is allowed to do after closing.

In August 2023, e&, the UAE-controlled telecoms company, agreed to acquire 50% plus one share of PPF Telecom’s businesses in Bulgaria, Hungary, Serbia and Slovakia. The price was €2.15 billion, plus a potential €350 million earn-out.

The European Commission received the Foreign Subsidies Regulation notification in April 2024, opened an in-depth investigation in June and granted conditional approval on September 24. The parties disclosed that the transaction closed in October 2024, roughly 14 months after agreement.

The approval required commitments restricting financing from the Emirates Investment Authority and e& into the EU businesses. It also required market-term dealings and monitoring by an independent trustee.

Here is the twist: e& bought the asset, but not the unrestricted ability to fund it as originally expected. The gate reached beyond purchase price and into the future economics of ownership.

This is where founders and investors need to get painfully specific. Consider an illustrative example: an investor agrees a £10 million round, but regulatory approval is pending. The term sheet might make approval a condition before closing, give either party a long-stop date, require the company to retain enough cash to survive beyond that date, and allow the investor to walk away if approval conditions fundamentally restrict the investment.

Those four clauses change value. The headline valuation may remain £10 million. Yet the company has less certainty, management must protect a cash reserve, and the investor owns an option to leave if the regulatory outcome changes the economics. A lower valuation is only one way to price approval risk. Time and exit rights can do just as much work.

The paperwork that matters years later

The final gate may not appear at closing at all.

In 2018, Nigeria’s central bank alleged that MTN Nigeria had improperly repatriated $8.1 billion between 2007 and 2015 using problematic Certificates of Capital Importation. MTN said it had relied on commercial banks to obtain required approvals.

Additional documentation resolved most claims. But MTN paid a $52.6 million notional reversal relating to a 2008 private placement, while the central bank regularised certificates covering approximately $402.6 million.

A Certificate of Capital Importation is the bank-supported record that foreign capital entered Nigeria for a documented investment purpose and can later support lawful repatriation.

This is not filing for filing’s sake. It is exit protection.

The documents explaining why capital arrived can matter years later when dividends, sale proceeds or capital need to leave. Preserve authenticated wire records, final approvals, valuation evidence and capital-import certificates for the life of the investment.

Put the three cases side by side and the deeper pattern becomes clear. Cerebras shows that a changed security does not guarantee settlement. Great Wall shows that a timetable can be overtaken by politics. MTN shows that even completed funding can be challenged later if the documentary trail is weak.

The common mistake is thinking that cross-border capital has one finish line. It has several, and they occur at different times.

What founders should do on Monday

Separate four dates: signing, approval, wiring and share issuance. Do not call all four “closing”. Put each into the cash forecast separately.

Ask the investor for more than proof of funds. Ask what internal authority and outbound-transfer process it needs, who ultimately owns it and whether that ownership creates a screening issue.

Give both banks the payment purpose, ownership information, source-of-funds records and corporate documents before the planned wire date.

Prepare company-side mechanics early: board authority, share terms, valuations where required, sector checks, reports and local filings. If India is involved, work backwards from the 60-day allotment requirement.

Then negotiate for delay as seriously as you negotiate valuation. Define the approval condition. Set a realistic long-stop date. Decide whether cash must be reserved while approval is pending. Specify what happens if an authority permits the investment only with conditions that damage its economics.

These are the instincts behind the data room was 95% done, but lenders still would not fund: completeness is not the same as fundability.

What experienced investors see first

Investors are not being fussy when they probe ownership chains and payment routes. They are managing a precise risk: committing capital, reputation and time to a deal that cannot settle, cannot issue the promised security or cannot later return proceeds lawfully.

Experienced investors underwrite the path to cash, not just the business plan. They ask: who is the ultimate owner? Is the company in a screened sector? Can it issue this class of shares? Does the recipient bank know the transaction is coming? What proves the payment’s purpose? What happens after the long-stop date?

They also see the uncomfortable truth behind a headline valuation. A high valuation with an easy walk-away right, a short long-stop and no interim cash reserve may be less valuable to a founder than a lower-priced deal with a dependable route to close. The deal that dies in the investment committee memo is often the deal whose hidden condition was never raised in the pitch.

GI Network would build a capital-to-cash execution map for the specific investor, entity structure, country corridor and sector. That means identifying regulatory and banking gates before outreach, testing whether the proposed security can be issued, assembling the documents each bank will need, setting the critical path and rehearsing the objections likely to arise in an investment committee or regulatory review. Only then would GI Network classify a commitment as forecastable cash, conditional cash or commercial interest.

Use the Four-Gate Test

Before spending against a signed foreign commitment, run the Four-Gate Test:

1. 1.Investor gate: Can this investor legally and internally send the money, and is its ultimate ownership clear?
2. 2.Issuer gate: Can the company receive the money and issue the agreed shares on time?
3. 3.Regulator gate: Which approvals, sector restrictions or public-interest conditions must clear before funding or completion?
4. 4.Bank gate: Have the sending and receiving banks accepted the investor, source, purpose and payment documents?

For each gate, mark it cleared, pending with a dated path, or unknown.

Only cleared belongs in the bankable cash line. Pending belongs in a conditional scenario. Unknown is not a financing plan yet.

A signature may be an important moment. Across borders, it is often the moment the real work begins.