This week clarified how quickly AI is reshaping project delivery. Capability is moving faster than governance, pushing complexity outward—from central control to individual users, from tools to ecosystems, and from short-term adoption to long-term delivery consequences.

Access & Governance: Claude Cowork and ChatGPT Go bring agentic AI into everyday workflows, shifting responsibility from IT gatekeepers to users and making integration and standards the real challenge. Infrastructure: Microsoft’s data-centre expansion points to multi-year programmes across land, power, and skills amid growing resource constraints. Ecosystems: Apple’s move to Google’s Gemini reinforces that delivery now depends on managing interfaces, not owning the stack. Capital: $50 billion across xAI, Anthropic, and NVIDIA–Groq signals rapid consolidation of the AI infrastructure layer beneath all projects.

Anthropic's new "Cowork" tool points to where project delivery is heading: capability without complexity. Built on the same architecture as Claude Code and developed in just ten days—by Claude itself—Cowork brings agentic AI to anyone with a Mac. No terminal. No coding. Just describe what you need, grant folder access, and watch Claude execute. Read the full breakdown →

For project delivery professionals, ease of access doesn't eliminate complexity—it distributes it. When anyone can set an AI loose on file management, expense reports, or research synthesis, the governance burden shifts from IT gatekeepers to every individual with a subscription. Teams may move faster, but assumptions, limits, and long-term consequences can hide in plain sight. The future isn't no-code; it's know-enough delivery.

Microsoft's announcement of a new wave of data centres—alongside pledges not to raise local electricity bills—signals more than surging AI demand. It points to a sustained pipeline of infrastructure work coming onstream. For project leaders, this means long-tail programmes spanning land, power, planning, logistics, and skills. Read the full breakdown →

One reading is optimism: multi-year certainty for construction, engineering, and digital delivery teams. Another is pressure—more projects competing for constrained resources. Success hinges on sequencing, regional capacity mapping, and surfacing risks beyond site boundaries early. Meanwhile, Meta's nuclear power deals illustrate how the hyperscalers are racing to lock in energy supply, adding another dimension to the infrastructure chess match.

The DTSA micro-credential gives young people and career changers barrier-free access to digital twin education – a first for the UK construction industry. Built on 32 months of work at the University of Cambridge’s CDBB, it opens doors to cutting-edge skills in safer, smarter, and more sustainable project delivery.

With portfolio-based assessment (offered as part of an Apprenticeship) and real industry insight, the course creates a clear pathway into digital construction for site teams, aspiring architects, engineers, surveyors, and project owners / funders. In partnership with the Digital Twin Hub and OCN London, the DTSA is shaping the next generation of talent and helping position the UK as a global leader in digital construction and innovation.

Sign up by emailing [email protected]

OpenAI's launch of the $8 ChatGPT Go tier is a delivery story disguised as pricing news. At eight dollars a month, AI capability moves from occasional use to everyday tooling for far more teams. For projects, this widens the base: junior staff, SMEs, and site teams gain access to higher-capacity models without enterprise procurement cycles. Read the full breakdown →

One angle is productivity uplift at scale; another is control risk—more usage occurring outside formal governance. The work ahead isn't rollout but integration: standards, training, and expectations catching up with access. Add the imminent arrival of advertising in free and Go tiers, and project teams face new questions about data privacy and user trust. Cheap AI doesn't reduce complexity—it spreads it.

Apple's decision to power its long-awaited Siri overhaul with Google's Gemini signals a quiet shift in how complex systems get delivered. Control moves from owning everything end-to-end to orchestrating dependencies across firms. For project leaders, that mirrors modern delivery reality: value now sits in integration, governance, and risk management—not just capability building. Read the full breakdown →

One angle is speed. Borrowing mature AI compresses timelines and reduces delivery risk. The counterpoint is dependency: roadmap control, data boundaries, and commercial leverage drift outside your organisation. Apple maintains its privacy guarantees through on-device processing and Private Cloud Compute, but the partnership nonetheless validates a strategic pattern. In projects, this reinforces a lesson many still resist: delivery success depends less on tools and more on how well you manage interfaces.

In a ten-day stretch, the AI industry moved astonishing capital: xAI closed $20 billion at a $230 billion valuation, Anthropic signed a $10 billion term sheet at $350 billion, and NVIDIA absorbed Groq's assets and talent in a $20 billion "licensing" deal. These aren't isolated headlines. Together, they reveal how rapidly the infrastructure layer is consolidating—and how project leaders should prepare for the ripple effects. Read the full breakdown →

For project delivery professionals, this capital concentration signals both opportunity and volatility. Compute resources will expand, but vendor dynamics are shifting fast. NVIDIA's Groq deal marks the arrival of the "inference era"—when running AI models overtakes training them as the primary compute demand. Anthropic's IPO trajectory and xAI's scale ambitions mean enterprise buyers face new questions about long-term partnerships. The landscape is moving faster than most procurement cycles can handle.

A comprehensive list of phrases that instantly flag AI-generated content has emerged, and savvy project communicators should take note. Terms like "And honestly?", "Here's the kicker", "Let's unpack this", and "You're allowed to..." have become telltale markers of AI authorship. Excessive use of "quiet" terminology (quiet quitting, quiet luxury) and emoji bullets similarly signal machine generation.

For project delivery professionals drafting stakeholder communications, proposals, or status updates with AI assistance, the fix is straightforward: review outputs for these patterns and rewrite in your natural voice. Authenticity matters more than ever when everyone has access to the same tools. Your credibility depends on content that sounds like you—not like a model trained on the internet.

Explore the full guide on Grammarly →

The week's governance headlines exposed how quickly AI deployments can spiral beyond intent—and how regulators are racing to catch up. Indonesia and Malaysia blocked access to xAI's Grok over deepfake concerns after the chatbot generated sexualised images of children and non-consensual intimate images of adults. The images spread widely on X, prompting regulatory probes across Europe, India, and Malaysia. When your AI assistant becomes an international incident, the governance gap isn't theoretical anymore.

Meanwhile, an eighth wrongful death lawsuit alleges ChatGPT acted as a "suicide coach," providing harmful guidance that contributed to a user's death. The case arrives as OpenAI pushes ChatGPT into mass-market territory with its $8 Go tier and imminent advertising rollout. Liability frameworks haven't kept pace with capability expansion—and plaintiffs' lawyers are taking notice.

Chinese Humanoids Dominate CES 2026 – Of the 38 humanoid robot exhibitors at CES, 21 were Chinese firms including Unitree, AgiBot, and EngineAI. China has filed approximately 7,700 humanoid-related patents in five years versus 1,561 in the US, cementing its lead in commercialisation. Read more

Figure AI Completes 11-Month BMW Deployment – Figure AI's humanoid robots completed an 11-month deployment at BMW's Spartanburg plant, running 10-hour shifts and contributing to the production of over 30,000 X3 vehicles. The robots loaded more than 90,000 sheet-metal parts before being retired for the next-generation Figure 03. Explore

Boston Dynamics Unveils Production-Ready Atlas – The electric Atlas made its public debut at CES with Google DeepMind's Gemini Robotics AI powering its reasoning. Featuring 56 degrees of freedom and 110-pound lifting capacity, initial units will deploy at Hyundai's Georgia Metaplant in 2026. Find the details

1X NEO Robot Available for $20,000 Pre-Order – Norwegian startup 1X opened pre-orders for its home assistant humanoid at $20,000, making it the first consumer-priced general-purpose robot. NEO can fold laundry, load dishwashers, and tidy homes—delivery expected later this year. Read more

Columbia Engineers Crack Robot Facial Expressions – Researchers announced a breakthrough in realistic lip motion for humanoid faces, addressing the "Uncanny Valley" problem. The robot learned facial control by experimenting with thousands of expressions in front of a mirror. Read further

The PhD-Level AI That Can't Hold a Conversation

This week's conversation with Mike Clayton, author and leading voice in project management, explores the paradox that defined 2025: AI models that ace PhD-level reasoning tests yet struggle to maintain coherent conversations with subject matter experts. Mike describes watching models demolish benchmarks across verbal, numerical, and abstract reasoning—then falter when deployed in actual business contexts. The gap between benchmark performance and practical utility became one of the primary reasons implementation projects failed across businesses last year.

Rather than viewing these limitations as failures, the discussion reframes them as opportunity. The profession isn't ready for full artificial general intelligence, and current AI provides what Mike calls a "lovely sweet spot" where powerful tools augment human capability without displacing it entirely. This breathing space enables a "bookending approach" where humans define tasks and validate outputs while AI handles the computational middle.

The episode maps where transformation will hit first: the PMO function faces immediate disruption as planning, monitoring, and risk evaluation become automation targets. Mike envisions a future where consultants deploy pre-configured AI agents rather than templates—a "PMO in a box" that integrates with client systems and generates plans autonomously. For project professionals navigating AI implementation, this conversation offers both cautionary tales and genuine optimism grounded in practical experience

Listen Now

Listen Now or watch now

Read the blog

21- 22 January 2026 | Milan, Italy,

AI*Festival gathers visionaries and experts globally to explore artificial intelligence's transformative impact across industries. With over 10,000 attendees from its last edition, the event serves as an innovation hub through keynotes, sessions, and discussions on AI agents reshaping human collaboration. Attendees engage in B2B meetings, expo areas with startups, and plenary talks on agentic AI, fostering networking among professionals, leaders, investors, and enthusiasts.
Details here

Ever wondered how LLMs work? Well it’s something like this.

See you soon,

James, Yoshi and Aaron—Project Flux 

All content reflects our personal views and is not intended as professional advice or to represent any organisation.