The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.
A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.
Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail
AI that generates 3D models, scenes, and textures from text descriptions, images, or procedural rules. Includes text-to-3D pipelines and PBR material generation; distinct from game and AR/VR content which targets interactive rather than static 3D output.
AI-driven 3D asset generation has definitively achieved technical commodity status by June 2026, with full-pipeline automation now embedded in major platforms. The market inflection from research to production is complete: Tripo reaches 10M+ users and 90,000 studio clients with named customers (Sony, NetEase); Snap Lens Studio embeds native text-to-3D with Meshy partnership as first-class developer feature; Meta releases SAM 3D as open-source production-grade foundation model. Generation quality has crossed the production threshold—practitioners report 2026 output with automatic topology and retopology "drops directly into games," a stark contrast to 2024 workflows requiring extensive manual refinement. The vendor ecosystem demonstrates this maturity through convergence: open-source models (Hunyuan 3D, TRELLIS) now match proprietary quality while offering unit-economics advantages at scale; platform adoption metrics (TRELLIS 804.9K runs, Hunyuan 3D 21.7K runs post-launch, Meshy $40M ARR, Tripo $200M Series A+ funding at unicorn valuation) confirm production-scale usage. However, the practice remains constrained by organizational integration, not technical capability. The binding bottleneck has shifted from asset creation to asset packaging—raw output still lacks complete materials, metadata, and rigging for production integration. E-commerce deployments (Lowe's 30,000+ models at <$1 each, 2x conversion lift) show clear ROI in vertical markets. Horizontal enterprise adoption still lags despite technical sufficiency and vendor tooling maturity. The challenge is now organizational: workflow integration, business case clarity, and change management—70% of adoption failures remain non-technical. Practitioner assessments document remaining deployment friction: per-asset overhead (scale correction, collision setup, material adjustment, rigging) slows scaling despite generation speed improvements. Indie games, fashion, and e-commerce continue scaling; enterprise adoption remains pilot-bound. The practice has crossed from leading-edge innovation to commodity infrastructure, but production deployment at organizational scale awaits organizational alignment rather than technological breakthroughs.
June 2026 vendor maturity milestone: Full-pipeline platforms now operational with enterprise funding validation. Tripo (now Vast/INCE Capital-backed unicorn) reaches 10M users, 90,000 studio clients including Sony and NetEase, $200M Series A+ funding concurrent with product releases (H3.1 geometry, P1.0 production-ready meshes in seconds). Snap Lens Studio integrates native text-to-3D with Meshy partnership for PBR materials—major AR platform making 3D asset generation first-class developer feature. Meshy achieved $40M ARR and 10M+ cumulative users with 100M+ models generated. Tencent's Hunyuan 3D v3.1 became most-used 3D model on OpenRouter platform post-April launch, deployed across 150+ enterprise customers with 28k+ GitHub stars, 3M+ HuggingFace downloads, and game-art-standard topology (74% token reduction). Replicate platform metrics show TRELLIS at 804.9K runs and Hunyuan 3D at 21.7K runs—Replicate recommends Hunyuan 3D 3.1 as "best all-around model," signaling open-source convergence with proprietary solutions. Meta releases SAM 3D as open-source foundation model with CVPR Best Paper Honorable Mention recognition and 5:1 human evaluation win rate over competitors. Vendor ecosystem pricing: $0.40-0.50/run for high-end tools (Rodin, Tripo, Meshy); open-source models effectively free at scale for on-premise deployment. Unit economics have inverted—at 10k+ asset production volume, open-source (Hunyuan3D, TRELLIS) on consumer GPUs cost-favorable vs SaaS platforms.
Critical shift in production barriers: Quality is no longer bottleneck. Practitioners confirm 2026 output quality with automatic topology "drops directly into games" vs 2024 requiring weeks of refinement. Game development pipelines document concrete compression metrics: hero props 3-5 days → 4-6 hours; background assets 1 day → 30-45 minutes. Generation is now embedded across all major DCCs (Godot, Autodesk, Blender, Unity, Claude) via integrations. However, CTO-level analysis confirms asset packaging remains binding constraint—raw output still lacks complete materials, metadata, and rigging for production integration. Deployment friction analysis shows per-asset overhead (scale correction, collision shapes, material adjustment, rigging retargeting) remains significant scaling bottleneck despite generation speed. E-commerce shows clear ROI: Lowe's (30,000+ models <$1 each), IKEA, Cartier, Richemont report 2x conversion lift and 40% return reduction. Vertical markets (fashion, e-commerce, indie games) scaling at 19.5% CAGR; market forecast $3.23B (2026) growing 30.9% YoY to $9.4B by 2030. Emerging deployment verticals beyond games/media include automotive (component prototyping), manufacturing (3D printing workflows with automated segmentation and joint generation), and enterprise digital twins with infrastructure cost reduction (50% production time savings reported in field deployments). Enterprise horizontal adoption stalled: 95% of GenAI pilots fail to scale, 70% of barriers organisational (workflow integration, ROI clarity, change management). Production failure rates of 40-94% persist due to intent-loss and topology inconsistency in edge cases. Vendor viability risk: Stability AI exodus and $100M+ debt continues constraining model development investment. The paradox sharpens: technical commodity status with thriving vertical markets and major vendor investment, yet organizational adoption barriers dominate—workflow integration and business case clarity, not capability, now bind broader adoption.
— Games industry analyst acknowledges 3D generation reaching 'production inflection point' despite documented limitations: lacks scene assembly, animation limited to presets, no downstream lighting/physics tools—balances adoption momentum with honest capability gaps.
— Tripo v2.5 achieves <0.5 second generation from single image with integrated PBR textures and transparent usage-based pricing ($0.20-0.40/model), normalizing 3D generation as commodity service for e-commerce catalog automation and game prototyping.
— Independent company (not Meshy/Tripo) reports 49.4% YoY H1 2026 revenue growth driven by new AIGC 3D digital asset sales business, demonstrating market-wide adoption acceleration beyond single-vendor incumbents.
— WAIC 2026 conference (100k+ sqm, 1,100+ companies, 300+ launches) documents sector-wide pivot: 'Industry attention shifting from whether a model can be generated to whether the result can enter real production workflows,' indicating deployment-stage maturity consensus.
— Major game industry publication documents Meshy 3D Agent Beta launch showing production-grade conversational workflow integration with style consistency and animation-in-chat, advancing adoption from one-shot generation into professional studio iteration practices.
— Meshy Series B: $400M at $1.5B valuation with $40M ARR, 12M registered users, 100M+ models generated; product launches (Auto Split, Smart Topology, 8K Texture) and named customers (Nexon, NetEase, Square Enix, 37 Interactive) confirm production-ready ecosystem at unicorn scale.
— Hi3D deployment for mecha figurine 3D printing with multi-color segmentation and intelligent component splitting achieves 80% cycle-time reduction in industrial manufacturing, demonstrating quantified ROI and workflow integration in vertical market.
— Independent game-development practitioner blog documents real pipeline compression (hero props 3-5 days → 4-6 hours; background assets 1 day → 30-45 min) with honest assessment that AI generates 'first pass' requiring retopology and manual detail for hero-quality assets.