PLUS: Western scramble for AI answers, NVIDIA's biotech robotics secrets, and why 41% of shoppers now trust AI agents

Morning all, One big revelation coming out of CES 2026 is Alibaba's Qwen AI models now fuel over 63% of new AI derivatives on Hugging Face - a clear sign of China's accelerating dominance in open-source infrastructure.

AI will be the new frontier for geopolitics and this open-source trend could see Chinese firms controlling the foundational layer Western developers rely on. This raises some pressing questions: Can US-EU partnerships respond fast enough to prevent strategic dependence on Eastern AI ecosystems? Or will China control the AI infrastructure the same way they have manufacturing.

Today's dots:

• China's Qwen AI conquers Hugging Face landscape
• 41% of shoppers now trust AI shopping agents
• NVIDIA robotics slash cell therapy production costs
• Apple taps Google to rebuild Siri with Gemini
• Hidden productivity drain of unverified AI code

China's Open-Source AI Floodgates Open

Here's the thing: Alibaba's Qwen AI models now fuel 63% of new AI derivatives on Hugging Face, marking China's decisive lead in open-source innovation revealed at CES 2026.

Let's unpack that:

• Qwen models surpassed Meta's Llama as the most downloaded AI family on Hugging Face in 2025, spurring over 100,000 new custom variants
• The 'open-weight' approach lets developers tweak pre-trained parameters - adopted by robotics firms like Unitree for industrial automation and domestic helper bots
• Over 1,000 Chinese companies swarmed CES, with DeepSeek joining Alibaba in releasing high-performance open models for commercial use
• US-Asia partnerships accelerated at the event, including Dolby Labs teaming with Chinese manufacturers Hisense and TCL

If you remember nothing else: This shift shows Chinese firms mastering the global AI infrastructure layer Western startups rely on. Your next open-source project might default to Eastern foundations unless Western labs accelerate collaboration.

How AI Agents Are Quietly Taking Over Your Shopping Experience

Here's the thing: 41% of global shoppers now use AI assistants for product discovery according to a major IBM study, while enterprise tools like Aible's NVIDIA-powered retail agents are predicting volatile demand before humans spot the trend.

Let's unpack that:

• That 41% adoption figure marks a 62% surge in assistant usage since 2024 - with Baby Boomers leading growth at 92% (IBM). Younger groups use bots for bargain hunting while luxury shoppers leverage them for premium recommendations.
• Retailers like CVS and Verizon now deploy Aible’s NVIDIA DGX Spark-powered agents to detect micro-shifts in demand patterns across billions of transactions, with early adopters seeing 18-23% inventory waste reduction (Aible announcement).
• Behind the scenes, only 24% of consumers fully trust AI recommendations - the winning play sits in hybrid experiences where agents augment (not replace) existing channels like physical stores (cited in 72% of journeys).
• C-suite scramble alert: 54% of retail execs cite technical debt and data silos as blockers preventing them from moving beyond basic chatbots to transactional AI assistants.
• Watch these two 2026 battlegrounds: retailer-controlled recommendation engines (like Amazon's Rufus seeing 100%+ Black Friday conversion boosts) vs third-party LLM platforms monetising via sponsored product placements.

If you remember nothing else: Consumers are adopting AI shopping tools faster than businesses can integrate them. The winners will be retailers who treat agents as collaborative tools that enhance human decision-making - not replacements for it.

AI Robots Slash Cell Therapy Costs While Boosting Production

Here's the thing: Multiply Labs' new AI-powered robotic system is revolutionising cell therapy manufacturing - cutting production costs by 70% while achieving 100x greater throughput in contamination-free environments.

Let's unpack that:

• Their NVIDIA Isaac-powered robots now automate delicate cell modification processes that traditionally required human technicians in sterile 'bunny suits', reducing both contamination risks and labour costs
• Using NVIDIA Omniverse digital twins, researchers simulate thousands of manufacturing scenarios to identify optimal workflows before real-world deployment
• GR00T-powered humanoids handle material transport outside controlled environments, preventing contamination during loading/unloading - previously responsible for 38% of process failures
• Proprietary 'imitation learning' lets robots acquire skills by studying video demonstrations from top scientists, preserving critical institutional knowledge when experts retire
• This could slash current £100,000+ cell therapy prices to £25,000-35,000 per dose - making personalised cancer treatments accessible to millions

If you remember nothing else: These robotic clean rooms aren't just about efficiency - they're about reliably producing life-saving treatments at scale. As Multiply Labs deploys systems globally this year, personalised cell therapies may finally move from ultra-niche to mainstream medicine.

The AI 'Verification Bottleneck' Crushing Productivity

Here's the thing: AI now generates 42% of all code according to Sonar's latest developer survey, but 96% of engineers don't trust its outputs – creating a dangerous backlog of unverified code that could slow deployments and introduce security vulnerabilities.

Let's unpack that:

• 96% of developers don't fully trust AI-generated code yet 48% skip verification steps before committing it, creating critical technical debt according to Sonar's State of Code survey
• The verification backlog grows exponentially as AI-generated code requires 38% more review effort than human-written equivalents
• Productivity gains vanish as 24% of developers' work week now gets consumed reviewing and debugging AI outputs
• Resolve's operational analysis shows incident response capacity grew just 15% last year while code deployments increased 67% – the gap widening as more AI-generated features ship
• Next-gen AI operations tools like Resolve's system are emerging to integrate code, infrastructure and telemetry monitoring in a unified platform to manage this complexity

If you remember nothing else: Organisations celebrating coding speed gains from AI must urgently invest in verification systems and operational tooling. The bottleneck has shifted from creation to deployment confidence – ignore it at your peril.

The Shortlist

Target joined Walmart and Shopify in backing Google's UCP standard, creating a unified framework for agentic commerce. The move signals a major alignment in retail AI ecosystems, with OpenAI's Agentic Commerce Protocol facing competition.

UNESCO updated its AI Ethics Recommendation to address military applications, emphasizing human oversight. The framework now includes tools like Ethical Impact Assessments to audit autonomous weapons as member states grapple with accountability.