Inside Beijing’s Culture-Tech Roadshow: How AI Rewrote Cultural Production in 2026

The 2026 Beijing Culture Forum's cultural industry investors conference, held on September 23, delivered a decisive verdict on where capital and creativity now converge.

Culture-tech firms took the stage not as novelty acts but as serious commercial enterprises, showcasing artificial intelligence, digital twins, and spatial computing across heritage protection, video generation, and 3D reconstruction.

The roadshow signaled that technology has graduated from a supporting role into the very engine of content creation, product development, consumer experience, and monetization.

What emerged from the presentations was a portrait of an industry scaling with remarkable velocity. Companies founded barely three years ago now command tens of millions of registered users, generate hundreds of millions of videos, and export short dramas to more than two hundred countries.

Investors in the room scrutinized technology barriers, commercialization pathways, market space, and overseas expansion with equal rigor, treating cultural production as a discipline demanding defensible moats and repeatable revenue.

The deeper message is structural rather than anecdotal. China's cultural industry is being rebuilt around proprietary models, registered datasets, and platform economics, transforming heritage conservation, entertainment, and creative commerce into data-driven sectors.

This analysis dissects the roadshow's key players, their metrics, and the strategic logic that now governs cultural technology investment.

TL;DR At the 2026 Beijing Culture Forum roadshow, six culture-tech firms demonstrated how AI, digital twins, and spatial computing have become core production infrastructure for cultural industries. Jianwei Libao applies 3D scanning and IoT to preventive heritage protection across 2,000+ projects, while Shanhai Xingchen's NOVA AI powers short dramas reaching nearly 10 million monthly users in 200+ countries. KIRI Innovation, Dishi Technology, Pixel Bloom, and Aishi Technology collectively show massive scale: 126.7 million registered users for PixVerse, 30 million for AiPPT, and 200 million yuan in confirmed 2025 revenue. The event confirms that technology barriers, commercialization discipline, and overseas expansion now define competitive advantage in cultural production.

The Roadshow's Strategic Significance for Cultural Technology

The Beijing Culture Forum roadshow functioned as a capital-market audition, where culture-tech firms presented metrics designed to withstand investor scrutiny. Each founder faced questions on technology barriers, commercialization viability, market space, and overseas expansion potential.

The format itself reveals how cultural production has matured into an investable asset class demanding quantifiable defensibility.

What distinguished this cohort was the fusion of cultural domain expertise with proprietary technical infrastructure. Heritage firms deploy IoT sensors and digital twins; entertainment companies run self-developed AI production tools; visual platforms register image models and accumulate patents.

This convergence means cultural enterprises now compete on engineering depth as much as creative output.

The geographic and temporal context matters enormously. A September 2026 forum in Beijing, convening investors around Chinese cultural exports, reflects a deliberate national strategy of cultural出海 — culture going global through technological leverage.

The roadshow's participants collectively reach over 170 countries, embedding Chinese creative software into global workflows.

Understanding this event requires abandoning the assumption that culture and technology occupy separate spheres. The presentations demonstrated that AI now writes, renders, reconstructs, and distributes cultural content at industrial scale, while investors evaluate these capabilities through the same lens applied to any high-growth technology sector.

From Supporting Role to Core Engine

Technology's migration from backstage tooling to frontstage creation represents the roadshow's central narrative arc. Earlier cultural digitization meant scanning artifacts or archiving footage; today's firms generate original content, diagnose structural damage, and commercialize intellectual property through AI-native pipelines. The supporting role has ended.

Jianwei Libao exemplifies this shift by moving heritage work from rescue repairs toward preventive protection. Its 3D scanning, digital twins, and IoT monitoring create continuous data streams that predict deterioration before damage occurs. Prevention replaces reaction as the operative philosophy.

Shanhai Xingchen applies the same logic to entertainment, running all productions through its self-developed NOVA AI tool since early 2026. Content creation becomes a software function, with monthly capacity exceeding one hundred short dramas and box office surpassing 400 million yuan.

This transition demands new organizational competencies. Firms must maintain model training pipelines, data governance frameworks, and creator ecosystems simultaneously. The roadshow's investors clearly weighted these operational capabilities as heavily as any creative track record.

Investor Criteria Reshaping the Sector

Investors at the conference evaluated four dimensions: technology barriers, commercialization, market space, and overseas expansion. These criteria impose discipline on a sector historically valued for cultural prestige rather than defensible economics. The evaluation framework itself signals maturation.

Technology barriers now mean registered models, patent portfolios, and proprietary datasets. Dishi Technology cites more than 200 patents and a registered image model called Hetu, while Jianwei Libao is building a heritage language dataset and an industry model named Wenbao Tianshu.

Commercialization evidence has become non-negotiable. Aishi Technology reported 200 million yuan in confirmed revenue for 2025 after beginning monetization in late 2024, demonstrating that AI video platforms can convert massive user bases into actual income.

Market space and overseas expansion intertwine throughout the cohort. Stardust TV covers more than 200 countries in 14 languages; PixVerse operates across 170-plus countries; Genki AI products reach over 100 nations. Global reach is now baseline expectation.

Heritage Protection as a Data Discipline

Jianwei Libao's presentation reframed cultural heritage protection as a data-intensive engineering discipline rather than a craft-based preservation activity. Founded in 2009, the firm has participated in 20 World Heritage projects and more than 160 national key cultural relic sites. Its portfolio spans over 2,000 projects across 29 provinces.

The technical stack combines 3D scanning, digital twins, and internet-of-things monitoring to record and continuously observe heritage structures. This architecture enables a philosophical shift from rescue repairs — intervening after damage — toward preventive protection that anticipates deterioration through continuous data analysis.

Central to this capability is a heritage language dataset and an industry model named Wenbao Tianshu. The model supports damage diagnosis, risk prediction, engineering support, inspections, and emergency response, effectively encoding conservation expertise into reusable machine intelligence.

Such an approach transforms heritage conservation economics. Instead of one-off restoration contracts, firms can offer subscription monitoring, predictive maintenance, and emergency readiness — recurring revenue streams that appeal strongly to investors evaluating commercialization potential.

Short-Drama Export and AI Production Pipelines

Shanhai Xingchen, founded in 2023, demonstrated how rapidly AI-native entertainment companies can achieve global scale. Its overseas app Stardust TV covers more than 200 countries and regions, supports 14 languages, and maintains nearly 10 million monthly active users. These figures represent extraordinary reach for a three-year-old enterprise.

Since early 2026, the company has used its self-developed NOVA AI tool for all productions. Monthly box office exceeds 400 million yuan, with more than 12,000 short dramas online and monthly production capacity above 100 titles. AI has become the production line itself.

Expansion plans reveal the platform logic driving cultural exports. Shanhai Xingchen intends to scale AI PGC production, open a UGC creator platform, and extend its AI content tools into entertainment, e-commerce materials, and digital-human livestreaming. Each extension multiplies monetization surfaces.

The short-drama format suits this model perfectly. Episodes are brief, serialized, and algorithmically distributable, allowing rapid iteration based on viewer response. AI compresses production cycles while multilingual support removes the localization bottleneck that traditionally slowed cultural exports.

Spatial Computing and 3D Reconstruction at Scale

KIRI Innovation, founded in 2019, occupies the spatial computing layer of the cultural technology stack. Its KIRI Engine focuses on pure-visual 3D reconstruction and serves more than two million professional creators. The platform converts ordinary photographs into usable three-dimensional assets without specialized hardware.

Its consumer application Remy demonstrates the velocity achievable when reconstruction technology meets accessible design. Launched first on HarmonyOS 6, Remy reached one million users in nine days and 2.8 million in roughly ten months, sustaining nearly one million monthly active users. Adoption curves like these validate consumer appetite for spatial content.

The planned addition of 4D reconstruction — capturing motion through time alongside three-dimensional form — signals where the sector heads next. Cultural documentation, virtual production, and immersive commerce all require dynamic spatial data, not merely static models.

For investors, KIRI's dual strategy of professional tooling and consumer applications offers two distinct revenue pathways. Professional creators pay for capability; consumers generate network effects and data that improve reconstruction quality across the entire platform.

Patents, Registered Models, and IP Moats

Dishi Technology, founded in 2016, builds its competitive position on intellectual property depth. The company reports more than 200 patents and a registered image model named Hetu.

In an era when model capabilities commoditize quickly, registered assets and patent portfolios constitute durable differentiation.

Its Genki AI platform spans the full arc from IP creation to sales, encompassing hundreds of intellectual properties whose products reach more than 100 countries.

This vertical integration means Dishi captures value at every stage rather than ceding margin to intermediaries.

The registered image model deserves particular attention. Registration creates legal and technical defensibility, allowing the firm to license, enforce, and build derivative services atop a protected foundation. Competitors cannot simply replicate what is formally registered.

Investors consistently probe such barriers because they determine long-term pricing power. A cultural technology company without proprietary models risks becoming a thin application layer, easily displaced when platform owners absorb its functionality.

AI Visual Platforms and Enterprise Adoption

Pixel Bloom, founded in 2018, operates the AI visual platform AiPPT with more than 30 million registered users. Its client roster spans consumers, businesses, and government entities, including Lenovo, Huawei, Honor, CITIC Securities, Amway, and Nongfu Spring. Enterprise validation of this breadth is rare.

The company's marketing agent, Xiaofang Tongxue, handles research, analysis, strategy, and content generation. This positions Pixel Bloom not merely as a design tool but as an autonomous marketing function, compressing workflows that previously required multiple specialized roles.

Enterprise adoption signals something crucial about cultural technology maturity. When major corporations embed AI visual platforms into routine operations, the technology ceases to be experimental and becomes infrastructure. Infrastructure contracts are sticky, recurring, and defensible.

The government client segment adds another dimension. Public-sector adoption implies compliance, security, and reliability standards have been met — hurdles that eliminate many smaller competitors and entrench established platforms.

AI Video Generation and Commercialization Proof

Aishi Technology, founded in 2023, delivered perhaps the roadshow's most striking growth narrative through its AI video platform PixVerse. By the end of 2025, the platform had accumulated 126.7 million registered users, a 564 percent increase, and generated 730 million videos — roughly fourteen times more than the prior period. Operations span more than 170 countries.

Commercialization began in late 2024, and by 2025 the company reported 200 million yuan in confirmed revenue. This figure matters enormously because it demonstrates that AI video generation can convert enormous free user bases into paying customers at meaningful scale. Confirmed revenue, not projections, anchors investor confidence.

The growth mathematics deserve scrutiny. A 564 percent user increase alongside a fourteen-fold rise in generated videos implies deepening engagement per user, not merely acquisition.

Users who generate more content become candidates for premium tiers, watermark removal, and commercial licensing.

PixVerse's trajectory also illustrates the global nature of Chinese AI video platforms. Operating across 170-plus countries means the platform competes directly with Western counterparts while benefiting from China's dense ecosystem of AI talent, compute infrastructure, and creator communities.

The Economics of AI Content Platforms

AI content platforms exhibit distinctive economics that investors at the roadshow clearly understood. Marginal cost per generated video approaches zero once models are trained, while storage, inference, and moderation create variable expenses. Scale therefore dramatically improves unit economics.

Consider the relationship between registered users and revenue. If PixVerse converts even a small fraction of 126.7 million users into subscribers, the resulting recurring revenue compounds rapidly. The platform's challenge lies in conversion optimization rather than demand generation.

Content volume also creates data advantages. Every generated video provides feedback that can improve model quality, which attracts more users, which generates more data. This flywheel, when functioning, produces winner-take-most dynamics within categories.

The formula governing platform value can be expressed simply, where ##V## represents valuation, ##U## active users, ##C## conversion rate, and ##R## average revenue per paying user:

This relationship explains why investors pressed on retention and conversion rather than raw registration totals. A platform with modest users but high conversion and low churn can outvalue one with enormous but disengaged audiences.

Overseas Expansion as Competitive Necessity

Every participant at the roadshow framed overseas expansion as central rather than optional. Stardust TV reaches 200-plus countries, PixVerse operates in 170-plus, Genki AI products sell in over 100, and KIRI's creator base spans the globe. Domestic markets alone cannot sustain these growth rates.

Cultural exports through technology differ fundamentally from traditional media exports. When a platform provides creation tools rather than finished content, it embeds itself in foreign creators' workflows.

Switching costs rise, and the platform becomes infrastructure rather than a disposable entertainment source.

Language support operationalizes this strategy. Stardust TV's fourteen languages and PixVerse's global availability mean creators work in native tongues while the underlying models handle generation. Localization ceases to be a bottleneck and becomes a feature.

Regulatory complexity accompanies this expansion. Different jurisdictions impose varying rules on AI-generated content, data residency, and cultural content. Firms that build compliance into their platforms early gain durable advantages over those retrofitting later.

Investment Logic and the Future of Cultural Technology

The roadshow's investor questioning revealed a coherent evaluation framework applicable across the entire culture-tech sector. Technology barriers, commercialization evidence, market space, and overseas expansion form four pillars that separate fundable enterprises from interesting experiments. Each participant was measured against all four simultaneously.

Technology barriers manifest differently by subsector. Heritage firms accumulate datasets and industry models; entertainment companies build production AI and creator platforms; visual platforms register image models and patents.

Regardless of form, the underlying question remains whether competitors can replicate the capability quickly.

Commercialization evidence increasingly means confirmed revenue rather than user counts alone. PixVerse's 200 million yuan and Shanhai Xingchen's 400 million yuan monthly box office provide concrete proof. Investors have learned that engagement without monetization is a story, not a business.

Market space and overseas expansion complete the framework. Domestic saturation looms in several categories, making international reach the primary growth vector. Platforms serving 100 to 200 countries demonstrate that cultural products can achieve global distribution through software.

Comparative Commercialization Benchmarks

Placing the participants side by side reveals instructive variation in how each converts capability into revenue. Heritage protection monetizes through project contracts and monitoring subscriptions; short dramas through box office and platform fees; visual tools through enterprise licensing and consumer subscriptions.

This diversity suggests the cultural technology sector will not consolidate around a single business model. Instead, multiple viable pathways coexist, each suited to different content types, customer segments, and regulatory environments. Investors can diversify across models rather than betting on one.

Enterprise-facing platforms like AiPPT exhibit longer sales cycles but stickier contracts. Consumer platforms like PixVerse scale faster but face higher churn. Heritage monitoring occupies a middle ground with government and institutional clients providing stability.

Understanding these benchmarks helps explain why the roadshow included such varied participants. A portfolio spanning heritage, entertainment, tools, and video generation hedges against category-specific risks while capturing the sector's overall expansion.

Risks Confronting the Sector

Optimism surrounding cultural technology must be tempered by genuine risks that investors at the roadshow surely weighed. Model commoditization threatens firms whose differentiation rests on capabilities that open-source communities can replicate within months. Registered models and patents mitigate but do not eliminate this exposure.

Regulatory uncertainty compounds technical risk. AI-generated content faces evolving rules on disclosure, copyright, and cultural sensitivity across jurisdictions. A platform operating in 170 countries must navigate dozens of regulatory regimes, each capable of disrupting operations.

Content moderation at scale presents operational hazards. Platforms generating hundreds of millions of videos cannot manually review output, yet failures can trigger platform bans, legal liability, and reputational damage. Automated moderation systems become critical infrastructure.

Finally, overseas expansion carries geopolitical risk. Cultural exports can become entangled in broader diplomatic tensions, leading to market access restrictions or consumer backlash. Diversification across many countries provides some insulation but not immunity.

What the Roadshow Signals for 2027 and Beyond

The September 2026 roadshow offers a forward-looking signal about where cultural technology investment flows next. Expect continued convergence between heritage, entertainment, and commerce as platforms extend capabilities across domains. Shanhai Xingchen's move into e-commerce materials and digital-human livestreaming previews this trajectory.

4D reconstruction, announced by KIRI Innovation, hints at immersive applications requiring dynamic spatial data. Virtual production, augmented reality commerce, and interactive cultural experiences all depend on such capabilities. The firms building this infrastructure today position themselves for tomorrow's applications.

Commercialization discipline will intensify. Investors rewarded confirmed revenue in 2025 and will likely demand profitability or clear paths toward it in subsequent years. User growth alone will no longer suffice as a fundraising narrative.

Cultural出海 — culture going global — will remain the dominant strategic theme. Chinese cultural technology firms have demonstrated they can compete internationally on product quality and scale.

The next phase involves deepening those positions and defending them against both Western competitors and emerging rivals from other regions.

The 2026 Beijing Culture Forum roadshow ultimately documented a sector in transition from cultural prestige to technological substance. Six firms, spanning heritage conservation to AI video generation, demonstrated that cultural production now runs on proprietary models, registered datasets, and platform economics.

The metrics they presented — 126.7 million users, 200 million yuan in confirmed revenue, 2,000 heritage projects — constitute evidence rather than aspiration.

For investors, the framework is now clear: technology barriers, commercialization proof, market space, and overseas expansion. For cultural institutions, the message is equally direct: technology has become the medium through which culture is preserved, created, and exported. The roadshow did not predict this future; it simply recorded its arrival.