Meshy has released Meshy 7, a new image-to-3D foundation model that prioritises alignment between user-provided images and generated 3D results. The model leads competitors across three alignment metrics: 81.0% on overall proportion, 79.7% on spatial distribution, and 59.8% on surface details, measured against reference 3D models excluded from training. Meshy 7's largest advantage appears in surface details, where it leads the nearest competitor by 5.3 percentage points. The model achieves alignment through an upgraded image encoder that reads inputs at multiple scales and higher resolution, rebuilt training data with exact geometry correspondence, and geometry alignment tracked during each training cycle. Meshy evaluates alignment by comparing generated 3D geometry directly against reference models, rather than using human preference or vision-language judges. Meshy 7 launched on 10 August and is available across all Meshy subscription tiers.

AI funding heats up as crypto awaits its regulatory green light. This week overview (july 20 - july 24, 2026): - Meshy raised nearly $400M Series B at a $1.5B valuation, the largest round to date for a company built specifically on AI-powered 3D generation, led by IDG Capital, Matrix Partners China, and Monolith Management. - Augustus raised $180M Series B led by Tiger Global at a $1B valuation to build a federally chartered "Global Dollar Bank" giving international fintechs stablecoin-and-SWIFT access to USD, on the back of a fresh OCC conditional national bank charter - total raised now $210M. - Glow emerged from stealth with $180M Series A (Sequoia, Cyberstarts, Greenoaks, Redpoint, Index) at a $1.2B valuation, building AI-native endpoint security for a world where AI agents sit on the corporate laptop. - Movement Labs, the team behind the Movement Layer-2 network, filed for Chapter 11 bankruptcy, reported this week, after its MOVE token collapsed roughly 94-99% following a market-maker token-distribution scandal that got it delisted from major exchanges. Meshy: the largest check AI 3D generation has ever seen. Three-dimensional content has been the industry's obvious-but-unsolved AI application for years: every other modality (text, image, video, voice) got its foundation-model moment, and 3D lagged because the data, the tooling, and the use cases were messier. This week that gap closed with the single largest funding event AI 3D generation has recorded. Meshy closed nearly $400M in Series B funding at a $1.5B valuation, its first publicly disclosed number, led by IDG Capital, Matrix Partners China, and Monolith Management, with existing shareholders Granite Asia, HongShan, BAI Capital, and Source Code Capital all returning. The company builds foundation models that generate print-ready, game-ready 3D assets directly from text, images, or sketches - collapsing what used to be a multi-day modeling workflow requiring specialized software and specialized humans into a prompt. The traction numbers explain why investors leaned in at this size: annual recurring revenue is growing roughly 12x year over year, the platform has more than 12 million registered users, and users have generated over 100 million models to date. Those aren't hobbyist numbers - they describe a tool already embedded in real production pipelines across gaming, e-commerce, and manufacturing, categories where a usable 3D asset has direct, measurable economic value the moment it's produced instead of weeks later. Alongside the round, Meshy shipped the "Meshy 3D Agent" - an agentic workflow that takes a rougher input (a sketch, a reference photo, a text description) and autonomously handles the multi-step process of turning it into a finished, usable model, rather than a single-shot generation the user has to iterate on by hand. Text and image generation got their infrastructure-scale rounds years ago; video followed. 3D was the modality everyone assumed would eventually get its moment once the underlying models were good enough - and this week's round is the market declaring that the technology has crossed that threshold, with real revenue and real usage to back the price tag rather than pure narrative. Augustus: a bank charter as the moat, stablecoins as the product. If Meshy is this week's proof that a new AI category just became investable, Augustus is proof that the oldest financial category - dollar banking - is being rebuilt with crypto-native rails underneath a very traditional wrapper. Augustus closed a $180M Series B led by Tiger Global, with Hummingbird, QED, and the founders of Nubank, Ramp, Circle, and Deel all participating, at a $1B valuation, bringing its total raised to $210M. The company is building what it calls the "Global Dollar Bank": a federally chartered U.S. bank giving international fintechs and banks direct USD account access and payment rails spanning SWIFT, ACH, SEPA, and stablecoin settlement. The real asset isn't the round size, it's the charter. In May, the Office of the Comptroller of the Currency granted Augustus preliminary conditional approval for a full national bank charter, making it only the eighth company since 2010 to receive that approval. Fintechs have spent a decade routing around the need for a charter through sponsor-bank partnerships; Augustus chose to become the bank instead, which lets it offer dollar access on its own terms. The use of funds tells you where the demand actually lives: expansion into Latin America, Southeast Asia, the Middle East, and Africa, regions where local fintechs and banks have the hardest time getting reliable, affordable USD rails, and where a stablecoin-backed dollar account solves an expensive, everyday problem rather than a speculative one. Augustus is the least "crypto-native"-looking story in this roundup, wrapping a stablecoin settlement layer inside the most conservative possible institution - and arguably the clearest evidence yet of what stablecoins are actually being used for in 2026. Glow: securing the endpoint now that AI agents live there too. Cybersecurity spending follows wherever the attack surface just moved, and this week the attack surface moved onto the AI agents now running on employee laptops. Glow emerged from stealth with a $180M Series A at a $1.2B valuation, an unusually large debut round backed by Sequoia Capital, Cyberstarts, Greenoaks, Redpoint Ventures, Index Ventures, and Lux Capital, among others. Founders Roi Tiger (former Meta VP of Engineering), Omer Singer (former Snowflake cybersecurity strategy lead), Ophir Arie (former Claroty VP of R&D), and Arnon Joseph (former Meta engineering leader) built the company around a single premise: endpoint security, as a category, was designed for a world where humans and static software occupied a device - not autonomous AI agents holding standing permissions and the ability to act on their own. "If you think of the past decade, everything was moving to the cloud and SaaS. Suddenly, AI lands on the endpoint" - Roi Tiger, Glow co-founder That prevention-first posture matters because the threat model flipped in both directions at once: AI tools give attackers a faster way to craft convincing, adaptive attacks, and AI agents given broad device permissions are themselves a new kind of insider risk if compromised or simply misconfigured. Traditional endpoint detection, built to flag known malware signatures, was never built to reason about an autonomous process with legitimate credentials doing something it was technically allowed to do. Glow already has paying customers in healthcare, retail, and financial services, sectors with the least tolerance for a security model built for the last decade's threats, and employs close to 100 people, 70% based in Israel. A $1.2B valuation on a stealth debut says Sequoia and company think "AI on the endpoint"; it's the default problem every enterprise is about to have. Movement Labs: what happens when a token scandal actually reaches bankruptcy court. Not every story in a VC roundup is a raise, some are the other end of the cycle, and this week crypto got a reminder of what that looks like when it goes all the way to bankruptcy court. MVMT Labs, the development team behind the Movement Layer-2 blockchain network, filed for Chapter 11 bankruptcy protection in the U.S. Bankruptcy Court for the District of Delaware, according to reporting this week. The filing reported liabilities of up to $10M against roughly $500K in assets, a startling gap for a project that had raised a $38M Series A led by Polychain Capital in 2024 and a roughly $100M Series B in early 2025 at a valuation near $3B. The collapse traces back to a controversial market-maker agreement around the initial distribution of 66 million MOVE tokens, which triggered mass liquidations and got the token banned from major exchanges including Binance. MOVE has since traded down as much as 94-99% from its highs, and the largest creditor in the bankruptcy filing is co-founder Rushi Manche himself, with a claim exceeding $1.6M. The court has set September 14, 2026 as the deadline for creditors to file claims, with a formal restructuring plan due by October 13, 2026. Why this belongs in a roundup alongside four funding rounds: Movement Labs raised from serious, credible investors (Polychain led its seed round) at a serious valuation, and still ended up in bankruptcy court within roughly two years - not because the underlying technology failed, but because of a token-distribution and market-making structure that blew up the community's trust before the product ever got a fair test.

Partner Content As game studios tighten budgets, AI's next test is consistency. Become a member of GB MAX to gain exclusive access to the industry and to the most influential global B2B leadership community in the business of gaming, entertainment, and tech. Join now and also get a VIP ticket to GamesBeat Next (Nov 2-3, SF). Presented by Meshy AI Keep Watching The pressure on game studios has increased in 2026. Teams are leaner, budgets remain tight, and players continue to expect richer worlds delivered faster. As AI moves deeper into production pipelines, studios are moving past the question of whether the technology works and toward a more important business challenge: whether it can deliver quality at scale. The question is no longer whether AI can generate a 3D model in seconds. It can. The real challenge is whether those assets are consistent enough to reduce production time instead of creating more work. For studios, the relevant metric is time-to-usable-asset: the time required to generate, refine, optimize and validate an asset in its target production environment. Meshy 3D Agent is designed around that workflow, maintaining creative context across multiple turns and converting selected concepts into exportable 3D assets. From faster assets to better workflows. The latest GDC State of the Game Industry survey shows AI adoption continues to grow, particularly in early-stage development. Among more than 2,300 industry professionals surveyed, 36% said they personally use generative AI tools in their work, with research and brainstorming (81%), daily tasks (47%), coding assistance (47%), and prototyping (35%) among the most common uses. Asset generation remains a smaller application, cited by 19% of respondents. However, concerns around quality and creative standards remain widespread. More than half of respondents said generative AI is having a negative impact on the game industry, with developers citing risks around job displacement, data sourcing, and the quality of AI-generated content. BCG's 2026 Global Gaming Report similarly found that AI adoption is expanding across the industry, with roughly half of studios using AI somewhere in their pipeline. Consistency at scale. Meshy says its 3D Agent lets developers establish an art direction once, refine it through conversation and generate related assets without restarting the creative brief for every object. Rather than treating each model as a separate prompt, the workflow is designed to keep an entire asset set visually coherent. Demo: Generating a Consistent 3D Game Asset Set For a game studio, consistency operates on three levels. The first is creative: assets need to share a recognizable visual language, with matching proportions, material treatment and level of detail. The second is technical - assets have to follow compatible polygon budgets, scale conventions, texture requirements and export formats. The third is workflow, meaning art direction and production context carry through from concept generation to refinement, rigging, animation and engine handoff. Meshy's strongest product claim sits on the first level: creative consistency across an asset set. The broader platform extends that workflow into remeshing, PBR texturing, rigging, animation and export, reducing friction between concept development and downstream production. From text prompt to game engine: Meshy's AI 3D workflow. Rather than replacing artists, conversational AI is finding its place earlier in production. Concept exploration, blockouts, secondary props, and prototyping can move faster, while hero assets and final creative decisions remain with development teams. Meshy's product pitch extends beyond generating a visually consistent batch. Within its broader game-asset workflow, teams can start from a text prompt, reference image, sketch or rough idea, then establish and refine a shared art direction through conversation. From there, the agent generates a batch of related concepts, and selected ones can be converted into 3D game assets. The workflow continues past generation: assets can be prepared to meet target polygon, topology and texture requirements, characters can be rigged and animated, and everything exports in standard formats for game engines and digital content creation tools. Meshy 3D Agent supports exports including FBX, OBJ, GLB, USDZ, STL, BLEND, 3MF and DXF. Those formats allow generated assets to move into workflows involving Unity, Unreal Engine, Godot, Blender, Maya, 3ds Max and Roblox. Demo: Moving an AI-Generated Character Further Into Production As of July 2026, Meshy says more than 12 million registered users have generated over 100 million 3D models through its platform, which SimilarWeb ranks as the top platform in its category by web traffic. The company has just closed a $400 million Series B round from a group of leading global investors, valuing it at $1.5 billion. Meshy has also recently added in-chat rigging and animation for generated characters, expanding its role further into the game asset pipeline. Consistency is the starting point, not the whole story. For game studios, consistent AI-generated 3D game assets are only the first proof point. Meshy 3D Agent connects that first step to the rest of the production pipeline. That broader workflow matters because speed at the first step has little value if teams lose the time again in cleanup and handoff. The more useful benchmark is whether AI shortens the distance between an initial brief and a usable in-engine asset. Meshy is betting that a conversational, human-in-the-loop workflow can do exactly that: automate repetitive generation and iteration while leaving final creative and technical control with artists.