Week of January 09, 2026 | Issue #2
🔥 THE BIG STORY
UK Fires the Starting Gun on ISSB-Aligned Sustainability Reporting
The UK's Financial Conduct Authority has launched its most significant sustainability reporting consultation in years. The headline: climate reporting goes mandatory, everything else gets "comply or explain."
Here's what you need to know:
What's Mandatory (from January 2027):
Climate disclosures aligned with ISSB S2
Scope 1 and 2 emissions
Climate-related risks, opportunities, and transition plans
What's Comply-or-Explain:
Scope 3 emissions (opt-out until 2028, then explain why not)
Non-climate sustainability topics under S1 (opt-out until 2029)
Why This Matters:
The FCA's approach is pragmatic. They're acknowledging what everyone in the trenches knows: Scope 3 data collection is still a mess for many companies. Rather than forcing bad data into reports, they're allowing time for systems to mature.
But here's the strategic angle: "Comply or explain" isn't a free pass. Investors will notice who explains and who discloses. Companies treating this as an excuse to do less will find themselves at a competitive disadvantage when capital allocation decisions are made.
The UK is also signaling alignment with the ISSB's "global baseline" approach, now covering approximately 60% of global GDP. For multinational companies, this is welcome news: one framework gaining traction beats the regulatory fragmentation we've seen in 2025.
What to Do Now:
Gap analysis: TCFD to S2 transition (most companies will find this manageable)
Scope 3 strategy: Decide if you're disclosing or explaining, and why
2027 readiness: That's your first mandatory reporting year
🤖 AI × SUSTAINABILITY
The Uncomfortable Truth: AI Is Driving a Gas Power Boom
While we debate AI's potential to solve climate challenges, here's what's actually happening: 2026 is on track to be a record year for new gas power projects worldwide.
According to Global Energy Monitor, the US has nearly tripled its gas-fired capacity under development. The driver? Data centers. More than one-third of new gas capacity is intended to directly supply AI infrastructure.
The numbers are staggering:
If all proposed US gas plants are built and operate their full lifetimes: 12.1 billion tonnes of CO₂
That's roughly double the US's current annual emissions from all sources
Global planned gas build-out: 53.2 billion tonnes of CO₂
The AI Sustainability Paradox:
We're in a peculiar moment. The same week that Anthropic released Opus 4.6 (spotting 500+ zero-day vulnerabilities in testing) and OpenAI launched GPT-5.3-Codex (raising "unprecedented cybersecurity concerns" according to Fortune), the infrastructure powering these models is locking in decades of fossil fuel dependency.
MIT estimates data center electricity consumption will approach 1,050 terawatt-hours by 2026 — that would make data centers the fifth-largest electricity consumer globally, between Japan and Russia.
What This Means for ESG Professionals:
If you're advising technology companies, expect scrutiny on:
Power Purchase Agreements (PPAs) for AI infrastructure
Scope 2 emissions transparency for cloud/compute services
Water consumption disclosures (data centers are thirsty)
The companies getting ahead of this — Microsoft's 366 MW solar PPA in Italy is a model — will differentiate themselves. Those hoping the issue stays under the radar will find regulators and investors increasingly interested.
AI Model Wars: What Sustainability Teams Should Know
This week saw the most intense 24 hours in AI history:
Anthropic's Opus 4.6 (Released Feb 5):
1 million token context window
New "agent teams" feature for autonomous task completion
Scored 72.7% on OSWorld (computer use benchmark)
Found 500+ zero-day security vulnerabilities during testing (per Axios)
OpenAI's GPT-5.3-Codex (Released within 30 minutes):
Designed for "agent-style development workflows"
Can operate a computer and complete end-to-end tasks
Released with unusually tight controls due to cybersecurity risks
Why This Matters for Sustainability:
These aren't just coding models anymore. They're autonomous agents that can research, analyze, and execute tasks across systems. For sustainability reporting:
Opportunity: Automated data collection across fragmented systems
Risk: AI-generated sustainability content without proper oversight
The Reuters analysis from last month remains relevant: as AI-generated reports proliferate, quality control becomes critical. AI augments, it doesn't replace. Your review processes need to evolve.
📋 REGULATORY & POLICY
Global Disclosure Landscape: Fragmentation Continues
Companies face an increasingly complicated 2026. The US and EU moved away from climate disclosure alignment last year:
US: Halted defense of SEC climate rule
EU: Retained CSRD/CSDDD but scaled back scope and delayed timelines
Meanwhile: ~40 jurisdictions aligning with ISSB (including UK, Australia, Hong Kong)
Yet investor demand persists. According to Morningstar Sustainalytics, 61% of asset owners view ESG regulations as helpful for standardising frameworks.
The Bottom Line: Companies maintaining disclosures and investments despite regulatory uncertainty are positioning themselves well. Those waiting for clarity may find the market has moved on without them.
GHG Protocol Releases Land Sector Standard
After five years of development with 300+ stakeholders and 96 pilot companies, the GHG Protocol has released its Land Sector and Removal (LSR) Standard — the first global standard for tracking agricultural emissions and CO2 removals.
Why it matters: Agriculture accounts for 22% of global emissions. Until now, companies lacked a credible methodology for reporting these impacts. The new standard, effective 2027, finally provides that framework.
If you're in food & beverage, retail, or any sector with significant agricultural supply chains, this should be on your radar.
📊 CORPORATE & FINANCE
xInsurers Lead on Climate Disclosure (Again)
ClimateWise's latest report shows the insurance industry continues outpacing other sectors on climate integration:
58% have conducted or plan to conduct double materiality assessments
39% integrating nature-related risks into board-level governance
One-third have submitted full transition plans; another third committed to developing them
Insurance has been the canary in the coal mine for climate risk. They're pricing it into premiums and building it into governance because they're on the hook when things go wrong. Other sectors would do well to follow their lead.
Hong Kong Broadens Taxonomy to Cover Transition Finance
Hong Kong's updated taxonomy now distinguishes Green, Transition, and Exclusion categories. This is significant for transition finance — enabling investors to price credible decarbonisation pathways for hard-to-abate sectors.
Transition bonds are forecast to be 2026's fastest-growing sustainable finance segment according to Moody's, with overall sustainable bond issuance stable at ~$900B.
💡 MARTIN'S TAKE
The AI Energy Reckoning Is Coming
This week crystallized something I've been watching for months: the collision between AI's promise and its environmental reality.
On one hand, we have genuinely transformative technology. Opus 4.6 finding 500 zero-day vulnerabilities isn't a gimmick — it's a preview of AI systems that can identify risks humans miss. Applied to sustainability, imagine models that spot supply chain emissions anomalies, identify greenwashing in competitor reports, or predict regulatory gaps before they become problems.
On the other hand, we're building this future on a foundation of gas power plants. The same companies championing AI-enabled sustainability are signing PPAs that lock in fossil fuel dependency for decades.
This isn't sustainable, and the market is starting to notice. The UK's new reporting requirements will eventually require disclosure of Scope 2 emissions from AI infrastructure. The companies getting ahead of this — investing in genuine renewable capacity, not just offsets — will be positioned well when that scrutiny arrives.
For sustainability professionals: start asking your technology providers hard questions about how your AI tools are powered. The answers might surprise you.
— Martin
That's all for this week. See you next Monday