MEGA Plans December AI Inference Marketplace Launch | Magica
MEGA Commits $30 Million to Products, With AI Inference Market Due in December
Editorial Team
••📖6 min read
The renamed MegaETH Foundation is funding its own products after an earlier shift away from its accelerator. Its December inference market must show how it improves on existing services that already route requests by price and reliability.
MEGA's $30 million treasury commitment covers several products, beginning with an AI inference market planned for December.
MegaETH had already announced a shift toward its own applications in July, alongside its accelerator's closure.
Existing inference services compete on price and reliability; MEGA has withheld its providers, models and market mechanics.
MEGA, the newly renamed MegaETH Foundation behind the Ethereum scaling network MegaETH, has committed $30 million from its treasury to new products, starting with an AI inference market planned for December. The funding gives substance to an earlier move toward building applications itself, but the announcement leaves the market's economics and operation undisclosed.
MEGA’s company-reported $30 million commitment covers several products; its inference allocation remains undisclosed. Source: The Block.
Treasury funding, with an unspecified inference budget
MEGA says all the new products will run on MegaETH, an Ethereum Layer 2 blockchain whose public mainnet launched earlier in 2026. The team says it has spent months working in stealth, according to the November 10 report on the commitment.
MegaETH co-founder Shuyao Kong, who outlined the team's shift toward its own applications in July, said in the report that the money comes from the existing treasury and covers a range of products. She described inference as the largest move, without specifying its share of the $30 million. This is an allocation of existing resources, not a newly announced financing round.
The team has raised more than $80 million across venture funding, community rounds and token sales. That includes a $20 million seed round led by venture investor Dragonfly in June 2024 and, according to Kong, about $40 million from the October 2025 public token sale after discounts for buyers accepting a one-year lock. Cumulative fundraising does not establish the treasury's current balance or how long the product budget will last.
Alibaba’s Qwen4-Max preview gives developers a fixed model to test through November 25. API access requires Token Plan, while claimed coding gains remain unquantified in the launch materials.
OpenRouter’s company-reported routing example deprioritizes a provider after outages. Prices are illustrative, not MEGA pricing. Source: OpenRouter.
Price routing already exists; interchangeability is the test
Inference means running a trained AI model to produce an output. MEGA says suppliers charge widely different prices for comparable inference, while buyers struggle to assess the risks of cheaper, unfamiliar providers. Its approach, called Purified Intelligence, aims to make inference from different suppliers more interchangeable so more can compete. Kong said developers are the intended customers.
Kong said in the report that MEGA has deliberately withheld the market's workings, participating providers and available models until launch. The announcement supplies no measured price reduction with which to test its claim.
OpenRouter, a managed service that routes requests to AI models and their hosting providers, already offers an alternative. In its routing guide, updated in September, it says it offers 400-plus models across 70-plus providers through one endpoint. For standard requests, its default provider selection favors low prices while deprioritizing providers with recent outages; it also offers fallback providers and price ceilings.
That establishes an existing capability, not proof that the two services will be equivalent. OpenRouter itself warns that providers can serve versions of a model at different numerical precision that perform differently. Its guide describes controls for precision, data retention, latency and throughput, and a quality-first routing step for requests using tools. It also notes that provider failover cannot protect against an outage of OpenRouter itself.
A meaningful savings comparison therefore needs the same model and comparable quality and service conditions. Selecting the cheapest supplier alone would not establish MEGA's promised interchangeability.
July’s accelerator closure accompanied a shift toward first-party applications. November’s announcement extended that strategy into AI inference. Source: The Block — Mega Mafia closure.
The product shift began before the rebrand
MEGA brings a record of developing a blockchain ecosystem, but the November announcement does not establish an operating record in inference. Its previous application strategy also helps explain why owning products matters.
In July, MegaETH announced the closure of Mega Mafia, its accelerator for applications on the network, after two years and two cohorts. According to the report on the closure, Kong said the program had supported about 20 teams that raised approximately $80 million in pre-seed through Series A funding. That money was raised by incubated teams, separate from MegaETH's own fundraising.
The program selected teams to live and build alongside the core team and offered technical, managerial and market-making support. MegaETH took no equity, ownership or governance positions in them. Kong said most of its successfully incubated applications were no longer being built with MegaETH and that little of the value had flowed back to the network.
The report identified departures including trading project Global Token Exchange's decision to build its own chain and attention market Noise's launch on Base. Kong said in July that resources would move toward first-party consumer applications, giving MegaETH direct relationships with users and responsibility for outcomes. November's inference commitment extends that earlier strategy into a developer-facing market.
The November report, citing DefiLlama, put the network's total value locked at about $12 million, down from a peak above $245 million. It also put the MEGA token near $0.05, more than 76% below its all-time high. These figures describe the blockchain's crypto economy; they do not measure demand for inference or the treasury's spending capacity.
MEGA says it has no equity owners and is aligned with the MEGA token, according to the announcement summary. The November funding report also describes a summer 2025 repurchase of a 4.75% stake comprising pre-seed investors' equity and token warrants. MEGA's mandate includes expanding token uses, but Kong did not specify its role in the inference market. The stated alignment leaves open how customers will pay and whether market activity will benefit token holders.
December must reveal the market and its dependencies
MegaETH says the network will remain general-purpose and permissionless, with continued technical, go-to-market and network support for ecosystem builders. Its launch announcement lists fast finality, sequencer rotation and proximity markets among the milestones still ahead as it pushes the chain further to support inference. Those are pending network milestones, not demonstrated inference performance.
December's launch needs to disclose which providers and models participate, how supplier quality is assessed, and what a request costs under comparable service conditions. It also needs to explain the blockchain's role and which upgrades are required at launch. Those details will let developers judge whether MEGA offers an economic or reliability advantage over services they can already use.
Railway estimates stolen customer keys generated about $740,000 in unauthorized AI spending after an October intrusion. Secrets from up to about 100,000 workspaces could have been decrypted, and the company says it will cover losses providers decline to refund.
Anthropic expects retail investors worldwide to receive up to 20% of its IPO, potentially $12.9 billion at the proposed upper price. Access through six brokerage channels depends on demand and broker rules, while a separate technology-company investment would help finance vast infrastructure commitments and founders and a special trust retain control.
Anthropic's seven founders plan equal IPO share sales worth about $250 million each at the proposed midpoint. Their sales make up 2.9% of the public offering, while the company seeks much larger financing under a structure that limits public investors' voting influence.
Nvidia’s independent directors retained Wachtell Lipton partner Joon H. Kim for an investigation into reports of a foreign intelligence effort targeting co-founder and CEO Jensen Huang. The board previously expressed confidence in Huang; the appointment provides no findings or timetable.
RRE Ventures led the funding for Graphon’s enterprise memory infrastructure. The startup is pitching connected, persistent context across different agents, but benchmark gains leave deployment costs and commercial scale unresolved.
Goldman Sachs, a lead bank on Anthropic’s IPO, reportedly forecasts $685 billion in annual revenue by 2030. The Claude developer has substantial cash reserves and rapidly growing sales, but rising compute costs, competitive pricing and delivery risks complicate the enterprise growth case.
Memories Upload pays contributors for accepted first-person video and sells annotated data to robotics developers. Its million-hour collection claim leaves open how much footage improves robot performance, while contributors surrender ownership without royalties.
OpenAI has acquired AI travel concierge Odessia and is bringing its founder, former Sonder chief Francis Davidson, into its travel effort. The standalone service has closed, leaving its future role alongside ChatGPT’s existing travel partners unclear.
Anthropic says expected advances from future models have triggered its research-acceleration threshold. An outside assessment finds faster progress but does not certify compliance with that threshold, while the voluntary policy leaves consequential decisions about safeguards and continued development with the company.
OpenAI’s free business-profile beta adds branded agents for eligible U.S. ecommerce and online service firms. The agents can also serve as paid-ad destinations, while OpenAI controls distribution and businesses bear responsibility for outputs.
Teleport has made Beams generally available with task-specific infrastructure permissions and a kill switch. Its access-security platform supplies an enterprise foothold, but established competitors and custom pricing for key controls complicate the buying decision.
New York will connect more than 80 researchers with Empire AI computing resources to evaluate advanced AI systems and advise state oversight, expanding a center first announced in January.
Anthropic’s amended IPO filing raises its Microsoft spending commitment while disclosing an exit right from a $34.9 billion Lambda agreement. Google’s larger commitment was already disclosed earlier, and expensive short-term computing continues to squeeze margins.
The cash deal would bring Oura’s existing AI and physician care partner under its ownership. The $159 annual integration already connects members to doctors, but geographic limits, extra service costs and unproven clinical benefits qualify the expansion.
Reddit is testing AI-organized review pages for roughly 10,000 products. The pages link back to users’ conversations, but lack purchase verification—a safeguard Amazon says it uses for its own AI review highlights.
Cambridge startup Tendrils Compute has raised a Benchmark-led seed round to build processors based on interaction nets. Its backers see an opening in branching workloads, including AI agents, but theoretical parallelism has yet to establish a commercial hardware advantage.
Google and Artificial Analysis launched a 200-task test of AI agents working with visually dense professional files. Google's study finds close leading scores and sharply different API bills, as its new Carbon model reaches customers alongside the Argon alternative.
Google is widening paid access to Gemini 4 Carbon for complex coding, business and scientific work. Gemini API token rates double after November 30, while benchmark results show that its cost advantage depends on the task and reasoning setting.
MiniMax has released downloadable weights for H3-Image, a model combining image generation and editing at native 16-megapixel resolution. Local deployment offers an alternative to its paid API, but hardware demands and a restrictive community license qualify the promise of open access.
