Tapas, Gnosis Pay, ZetaChain: a write-off, a pivot, and a migration (2026)

Between September 20 and September 27, 2026, three ventures announced the end of something they had spent years building. Kakao Entertainment said it plans to shut down Tapas, the North American webtoon platform it bought five years ago. 1 Gnosis Pay told its cardholders that their cards stop working on December 20. 2 ZetaChain's token holders voted to switch off the project's own blockchain and move the token somewhere else. 3

Each ending has a different shape. A parent wrote off an acquisition. A company retired a product and kept the business. A token community voted its own chain out of existence. What they share is where the deciding number sat: in what the user would do without the product, in which side of the business actually pays, and in which layer the money had been going.

Venture	Event and sector	Disclosed cause or evidence boundary	What happens to the people who depended on it	Counter-metric
Tapas	Kakao Entertainment announced on September 22, 2026 that it plans to shut down Tapas, its North American webtoon and webnovel platform. Publishers and rights holders were to be told the end date, content handling and settlements that day. 1 4	Kakao's stated reason is a refocusing on its Korean platform, Kakao Page. 5 Tapas itself never communicated the shutdown to its users, and no refund or user-data terms were announced. Trade reporting attributes the retreat to stalled growth in the United States. 1 4	Creators lose a place to host their work and their share of ad and currency revenue. Readers lose a storefront. Kakao keeps the IP and moves it to Kakao Page.	Owner-financed runway: the quarters the owner has actually committed, with any milestone-conditional money excluded
Gnosis Pay	Formal notice to all cardholders on September 21, 2026; the consumer debit card and web app stop functioning on December 20, 2026 at 23:59 UTC. Crypto debit card and payments. 2	No cause was given for closing the consumer product. The disclosed change is a repositioning that has been under way since 2025, from serving spenders to selling white-label card infrastructure to wallets, fintechs and neobanks. 2 6	Cards stop processing and ATM withdrawals stop. The web app stays for withdrawals. Cardholders are pointed to partner apps, including MiniPay, whose replacement cards run on Gnosis Pay's enterprise rails. Funds held in Gnosis Safe accounts are unaffected and have no deadline. GNO cashback ends September 30, with the final distribution on October 7. 2	Payer separation: the share of revenue from the party that signs the contract, counted apart from revenue from the party that uses the card
ZetaChain	Token holders approved Proposal 68 on September 20, 2026 with 99.4 percent support and 58 percent turnout, discontinuing the project's independent Layer 1 blockchain. ZETA migrates to Solana one-to-one as a native SPL token. Interoperability blockchain. 3	Disclosed: the team has pivoted from blockchain infrastructure to AI consumer applications. Its AI application, Anuma, launched in February 2026, passed 300,000 users and moves to Solana with the token. The exact date the chain stops has not been set. 3	Validators may keep running nodes until the migration finishes. ZETA held on Ethereum or BNB Chain is not part of the migration, and vesting schedules stay intact. A second proposal will set the snapshot height, the claims process and the timeline. 3	Raise-to-usage alignment: the share of active users and revenue in the product the last round financed, against the product users adopted on their own

1. Tapas: the platform a parent stopped paying for

The event

Kakao Entertainment announced on Tuesday, September 22, 2026 that it plans to shut down Tapas, the North American webtoon and webnovel platform it acquired in 2021. 1 Korean outlet Money Today reported that the company would notify publishers and rights holders that day with the service's end date, the handling of content and the settlement terms. 1 Kakao published nothing about user refunds, and nothing about what happens to reader data. 1

The price of the entry is on the record. Kakao acquired Tapas and the webnovel platform Radish in 2021, valuing them at about US$510 million and US$440 million. 1 The two merged into Tapas Entertainment in 2022, and Radish was closed in 2025, when Kakao said it would concentrate its North American story business on Tapas. 1 The trade outlet BigGo Finance puts the 2021 outlay at roughly ₩1.1 trillion, about US$810 million, and describes that investment as effectively written off five years later. 4

On the same day, Kakao Entertainment announced it would merge its two Korean story platforms, Kakao Page and Kakao Webtoon, into one service centred on Kakao Page before the end of the year, pooling about 16,000 original titles. 1 5

What the record supports

Kakao's own announcement of that consolidation describes it as part of organising the business around its core strengths so it can push harder abroad, and it does not mention Tapas at all. 5 The Tapas decision reached the platform's own users through Korean press reports first. Kotaku reported that readers and creators were shaken partly because Tapas itself had said nothing. 7 On that platform, writers earned a cut of advertising revenue and of the currency readers spend to unlock stories, which means a shutdown interrupts an income stream rather than only a reading habit. 7

The closest thing to an operating explanation is trade reporting. The Beat, quoted by The Daily Cartoonist, attributes the withdrawal to stalled growth in the United States, citing repetitive content among the causes, and notes that rumours of a shutdown had been circulating. 4 That is a newsroom's reading. The company has published no cause.

Read against those facts, the failure is a mistimed market entry: Kakao bought outright platform ownership in North America in 2021, at a price that assumed the market would keep expanding, and closed it five years later while concentrating on a Korean library it already owned. Waiting five years before giving up is the part worth studying, because the signals were internal long before they were public. A platform whose own parent stops mentioning it in the announcement of its growth strategy is a platform whose renewal case has already failed.

The counter-metric

For anything inside a larger owner — an acquirer, a parent company, a strategic investor — measure the owner's patience directly:

owner-financed runway: the number of quarters the owner has already committed money for, with every milestone-conditional amount removed from that count;
the milestone that resets the clock, stated as a number, and the date it is due;
the alternative the owner already owns, which is what your business will be compared against when that date arrives.

Kakao Page and its roughly 16,000 titles were that alternative. A platform being evaluated against a library the parent already holds has to prove that it adds readers the library cannot reach, and it has to prove it inside the owner's patience, not the founder's.

2. Gnosis Pay: the paying customer was never the spender

The event

Gnosis Pay sent a formal notice to all of its cardholders on September 21, 2026. The direct-to-consumer debit card and the web application stop functioning on December 20, 2026 at 23:59 UTC. 2 After that date the cards stop processing payments and ATM withdrawals. The web app survives in a reduced form, letting users pull out whatever balance remains. 2

The exit is organised. Cardholders who want to keep spending through the same rails are pointed at partner applications, including MiniPay, which handles the migration and issues replacement cards; those cards run on Gnosis Pay's business-to-business infrastructure. 2 Because balances sit in self-custodial Gnosis Safe accounts, users keep access to their funds with no withdrawal deadline. 2 The GNO cashback programme has a shorter clock: eligibility ends September 30, 2026, with the final distribution on October 7. 2

What the record supports

No cause was disclosed for closing the consumer product. What was disclosed is the direction of travel, and the company's own website states it plainly: Gnosis Pay now sells white-label card programmes to wallets, neobanks, crypto exchanges and payment processors, with contract terms of two years on its startup tier and three on enterprise, and the consumer card presented as something partner apps offer. 6 Reporting describes the same repositioning as having been under way since 2025. 2

The inference to be careful about is the reason. A consumer card programme carries costs that scale with every spender it adds — issuing, fraud, support, cashback — while the money from card programmes is usually made at the platform level, per programme and per corridor. The evidence here supports calling this a product mismatch between who used it and who paid for it, and it does not support a claim about a specific expense that broke. The company kept the technology, kept the customers that sign contracts, and retired the audience that does not.

For a founder, the useful part is what the shutdown did not touch. Users were not stranded and funds were not frozen, because the product was built on self-custodial accounts and the company had a partner to hand them to. A company that plans an exit for its consumer line before it needs one loses the transition cost instead of the customers.

The counter-metric

Split the revenue by who signs:

payer separation: the share of revenue from the party that signs the contract, measured separately from revenue from the party that uses the product;
cost to serve per consumer account, including issuing, support and any cashback, compared with the cost to serve one partner programme;
the migration you owe consumers if you close their product: who receives them, who issues the replacement, and what they keep.

Gnosis Pay's consumer card could not be judged on cardholder numbers. It had to be judged on whether a spender, paying consumer-card economics, was a profitable way to reach the fintechs that buy the rails.

3. ZetaChain: the chain was overhead for a different product

The event

ZetaChain's token holders approved Proposal 68 on September 20, 2026, discontinuing the project's independent Layer 1 blockchain and converting the ZETA token into a native Solana token. Support was 99.4 percent, turnout was 58 percent, and opposition and abstentions together accounted for 0.3 percent. 3 The migration converts tokens one-to-one, changes the decimal precision from 18 to 9 to match Solana's standard, mints nothing new and leaves total supply unchanged. It applies only to the native ZetaChain Layer 1 tokens; ZETA held on Ethereum or BNB Chain is outside it, and existing vesting schedules stay intact. 3

The date the chain actually stops has not been set. Validators can keep running nodes until the migration completes, and a second governance proposal is expected to fix the snapshot height, the claims process and the timeline. 3 The proposal first appeared on September 17, when ZETA traded at about $0.0342; by September 19 it was around $0.0400. 8 The same vote and turnout were reported by The Cryptonomist. 9

ZetaChain had raised $27 million to connect blockchains and launched its mainnet in January 2024, with investors including Blockchain.com and Jane Street Capital. 3

What the record supports

ZetaChain disclosed its reason, and the reason is about where its users were. The team pivoted from blockchain infrastructure to AI consumer applications. Its AI application, Anuma, launched in February 2026 and passed 300,000 users; Anuma migrates to Solana alongside the token. In that light, running an independent chain stopped looking like a competitive advantage and started looking like overhead. 3

The mismatch is worth stating precisely. The $27 million was raised for an interoperability network, and the demand that materialised was for a private AI assistant. What ended the chain was the gap between the $27 million and the users: the capital sat one layer below the demand. That is a product mismatch between the layer that was financed and the layer that was adopted, and the governance vote is the mechanism that let the project move without a liquidation.

The mechanics matter more than the headline. The 1:1 conversion preserves token count and supply, but the asset they hold is becoming exposure to a consumer AI product on someone else's chain, with the terminal date of the old network still unwritten.

The counter-metric

Sort usage and revenue by layer, not by company:

raise-to-usage alignment: the share of active users and revenue in the product the last round was raised against, against the share in the product users adopted without being asked;
cost of keeping the financed layer alive — validators, security work, maintenance — as a share of total spend;
what governance can decide, and whether your structure lets a pivot happen by vote rather than by liquidation.

Anuma's 300,000 users were visible months before the vote. A team with both numbers on one page would have seen which layer was paying for which.

Three counter-metrics to carry into the next review

Owner-financed runway: the quarters your owner has actually committed, with milestone-conditional money excluded, plus the date the renewing milestone is due.
Payer separation: the share of revenue from the party that signs the contract, kept apart from revenue from the party that uses the product.
Raise-to-usage alignment: the share of active users and revenue in the product your last round financed, measured against the product users adopted on their own.

Tapas, Gnosis Pay and ZetaChain ended in three different ways, and in all three the decisive number was one a founder could have written down beforehand: how long the owner's patience lasted, who was actually paying, and which layer the users had chosen.