MEGA Cuts 20% of Staff as $30M AI Product Push Reshapes Company

MEGA has cut 20% of its workforce. The reduction, disclosed Nov. 13 by co-founder Shuyao Kong in an X post, lands three days after the MegaETH Foundation rebranded and committed $30 million from its existing treasury to build products on MegaETH, starting with an AI inference market slated for December.

The company did not say how many people lost their jobs. There is no breakdown by team either. What we know is this: departing employees worked on the blockchain, wallet, ecosystem support and communications from the project's earliest days. Kong framed the cuts as preparation, saying MEGA had spent months getting ready to move in a new direction.

"We believe cryptography and distributed systems will change how the world produces and accesses purified intelligence," Kong wrote. She added: "We have a lot to prove, and a lot of work ahead."

The founders, early employees and core engineering team all remain, according to Kong. She thanked departing colleagues, called their contributions foundational to the company's next phase, and offered to introduce them to prospective employers.

What does the rebrand actually change?

Not as much as the name suggests. The MegaETH Foundation announced its new identity, MEGA, on Nov. 10 alongside the $30 million commitment. But the blockchain keeps its name. MEGA's official website still lists MegaETH as its blockchain division and describes MOSS as an account layer — one wallet across applications, with programmable permissions for both people and AI agents.

So the rebrand is organizational, not technical. MegaETH the chain, MOSS the account layer and now a product arm sit under one roof. The $30 million comes from treasury, not outside investors, which matters: no new capital entered, no new valuation was set.

How will the AI inference market work?

Inference means running a trained AI model to generate an answer or another output. MEGA says its proposed market would let customers source comparable services from a larger pool of providers than they can reach today.

The idea draws on a November white paper by Yilong Li and Lei Yang. They identify a trust problem: a cheaper service may not win customers who cannot establish which model is being used or how their data will be handled. Their proposed "Purified Intelligence" approach would make inference interchangeable across providers that users might otherwise have little reason to trust. The paper presents this as a mechanism for letting more suppliers compete for customer demand.

Kong told The Block that details on marketplace operations and available providers or models will stay under wraps until launch. All new products are intended to run on MegaETH, with planned network upgrades supporting the inference product.

Why the accelerator closed first

This week's moves follow a strategic pivot already underway. In July, crypto.news reported that MegaETH was closing its accelerator after two Mega Mafia cohorts supported roughly 20 teams that collectively raised $80 million. The program provided technical, management and market-making help without taking equity, governance rights or ownership in participating projects.

The results disappointed. Kong said most successful applications from the program were no longer being built on MegaETH. Global Token Exchange decided to build its own chain. Noise chose Base. HelloTrade moved toward Monad. Cap launched on MegaETH while pursuing a multichain strategy.

At the time, Kong said resources would shift toward first-party consumer products and "OMEGA" applications designed around capabilities the team considered specific to MegaETH. The inference market is the first concrete product of that shift.

The token and treasury backdrop

MegaETH launched its token on April 30, after 10 ecosystem applications met its first performance milestone. The project fixed supply at 10 billion MEGA, with 53.3% allocated to performance-based staking rewards.

In May, the foundation activated a stablecoin-funded buyback program. It completed its first purchase with net income generated by USDm's issuer through April 30. Future purchase sizes will depend on stablecoin supply and reserve yields.

USDm's reserves primarily used BlackRock's tokenized U.S. Treasury fund BUIDL through Securitize, alongside liquid stablecoins for redemptions. The foundation has said USDm was neither issued nor operated by the foundation or MegaLabs.

What does this change for MEGA?

The company is consolidating around a single bet: that crypto infrastructure can solve an AI trust problem, and that first-party products beat ecosystem subsidies. The 20% cut, the treasury commitment and the July accelerator closure all point the same way.

The December inference launch is now the test. If the marketplace delivers verifiable, interchangeable inference across providers, MEGA has a product story. If not, the company has cut its ecosystem teams and spent treasury capital on a thesis still waiting for proof.