In partnership with
Hey everyone, it’s Thursday, and FutureProof is back.
The title this week is: Oil Runs Short, EVs Go Global, and Google Reads Sign Language.
Oil supplies are tightening and prices remain volatile, making EVs, heat pumps and renewables look smarter by the day. Meanwhile, EV sales are surging across wildly different markets, solar is pushing coal aside in India, and Google has taught smartphones to understand sign language.
Highlights this week:
Climate: Regenerative agriculture reaches supermarket shelves, but vague definitions invite greenwashing.
AI: Europe starts enforcing its AI Act, Claude gets invisible watermarks, and Google translates sign language.
EVs: Denmark reaches 97% electric sales, China’s combustion market shrinks, and Tunisia quietly joins the surge.
Clean energy: Renewables squeeze European gas, Indian solar displaces midday coal, and Australia prepares to clean up its largest aluminium smelter.
Podcasts: Healthier soil becomes investible infrastructure, while supply-chain experts call time on decorative dashboards.
Yesterday’s solar eclipse crossed northern Spain. It wasn’t total where I went to see it in Cádiz, but it was still spectacular, and this edition’s image is an iPhone photo I took at its peak.
Right. Oil is running short, EVs are going global, and Google is reading sign language.
Let’s get into it.
Climate
Regenerative Agriculture Is Having Its Organic Moment, With One Big Catch
Regenerative agriculture is rapidly escaping the sustainability conference circuit and landing in actual shopping baskets: products carrying the label now generate roughly $2 billion a year in US retail sales, while consumer awareness has doubled in a remarkably short period. That’s encouraging, and particularly timely given this week’s Climate Confident conversation about soil becoming infrastructure, but there’s a snag: unlike organic, “regenerative” still lacks a consistent, federally enforced definition, leaving plenty of room for the marketing department to get ahead of the agronomy.
Sales are moving: Regenerative products span dairy, produce, snacks, baby food, tea and even pet food, with retail sales growing 10% over the past year after nearly 22% growth the year before.
Consumers are catching on: Awareness roughly doubled between 2023 and 2024, although only around 10% of US consumers actively seek regenerative products today, suggesting considerable headroom if brands can explain why they matter.
The label is the weak link: Certification schemes vary significantly, and unlike USDA Organic, companies can often make regenerative claims without independent verification. Greenwashing could squander consumer trust just as the category starts taking off.
Why This Matters: Regenerative agriculture is shifting from an environmental idea into a consumer, corporate and increasingly financial proposition, but credible measurement and standards will determine whether “regen” becomes the next organic or merely the next meaningless label slapped on expensive granola.
Kismet: The fascinating crossover with this week’s Climate Confident episode is that consumers and institutional investors appear to be discovering soil at almost exactly the same moment, one through the supermarket shelf, and the other through the balance sheet. 👉 Full story here
AI News
Europe’s AI Act Has Stopped Being a PowerPoint
Europe’s AI Act has crossed the line from legislation into enforcement: since 2 August, regulators can start policing general-purpose AI providers, banned AI practices and new transparency requirements. In other words, the era of nodding solemnly at “responsible AI” panels while quietly shipping whatever you fancy is becoming considerably more expensive.
AI now has to introduce itself: Chatbots and other interactive systems must make clear when users are dealing with AI rather than a human, while deepfakes must be labelled and synthetic content given machine-readable markings.
Frontier AI gets regulatory homework: General-purpose AI providers face documentation, copyright and training-data transparency requirements, while providers of the most powerful models must address systemic risks including cyberattacks, manipulation and loss of human control.
The penalties have teeth: The most serious breaches, including prohibited AI practices, can attract fines of up to €35 million or 7% of global annual turnover, whichever is higher. Suddenly “move fast and break things” requires a considerably larger legal budget.
Why This Matters: The EU has moved from writing the rulebook to actually refereeing the game, making AI governance an operational, engineering and board-level issue rather than another compliance document destined for SharePoint.
Kismet: More than 180 organisations have already signed the EU’s voluntary transparency Code of Practice, suggesting that, despite years of complaints about Brussels regulation, quite a few companies would rather help define the rules than discover them via a 7% fine. 👉 Full story here
Claude’s Words Now Come With an Invisible Passport
The EU AI Act has barely started biting and we’re already seeing product changes ripple far beyond Europe: Anthropic says new Claude models will embed invisible, machine-readable watermarks in generated text and attach signed provenance metadata to supported image files, worldwide. Brussels writes the rule, Silicon Valley changes the plumbing, and suddenly AI transparency isn’t theoretical anymore.
The watermark is meant to travel: Claude’s text markings are embedded at model level, so they follow output copied from the chat app, API, Claude Code and supported cloud platforms, and may survive some editing.
Images get provenance too: Supported files will carry signed C2PA metadata showing they passed through Claude, although screenshots and format conversion can strip that information away rather quickly.
It isn’t an AI detector: A human-written article merely proofread by Claude could carry a mark, while heavily rewritten AI text might lose one. Provenance is useful evidence; it is absolutely not a magic “bot wrote this” stamp.
Why This Matters: This is the Brussels Effect in unusually visible form, EU regulation is already changing how major AI systems behave globally, turning transparency requirements into actual product architecture rather than policy wallpaper.
Kismet: We may be heading for the wonderfully awkward situation where heavily edited AI prose looks “human”, while genuinely human writing that received a quick Claude polish carries an AI watermark, technology continues its proud tradition of making simple categories inconvenient. 👉 Full story here
Open AI Is Eating the Model Business
The AI model itself may be turning into a commodity far faster than expected: the best open-weight models are now snapping at the heels of the closed frontier, while open models increasingly dominate actual token usage in coding and agentic workloads. The intriguing bit isn’t that open AI is “catching up”, it’s that the economic value may already be migrating elsewhere, into agents, memory, orchestration and the systems wrapped around the model.
The capability gap is getting tiny: The report’s strongest open-weight model scored just four points behind the leading closed model on its benchmark composite, close enough that many businesses simply won’t need to pay frontier prices for everyday workloads.
Developers are voting with their tokens: Open-weight models now account for a majority of routed tokens on OpenRouter, and its seven highest-volume models in July were all open weight. Meanwhile, 79% of developers adding AI features say they use open models.
AI economics are getting weird, fast: Inference at roughly GPT-4-class performance has reportedly become about 50× cheaper in three years. If models become interchangeable and cheap, competitive advantage shifts towards the data, workflows, memory and agent infrastructure businesses actually control.
Why This Matters: If capable AI models become abundant commodities rather than scarce proprietary assets, the winning enterprise AI strategy may be less about picking “the best model” and more about owning everything valuable around it.
Kismet: There’s a geopolitical twist hiding inside all this openness: much of today’s open-model momentum is coming from China, meaning the technology that could loosen dependence on a handful of US AI giants may simultaneously expand Chinese influence over the global AI stack. 👉 Full story here
Google’s AI Can Now Read Sign Language
This is AI I can get properly excited about. Google DeepMind has built a sign-language-to-text model that lets Deaf users sign directly to their phone instead of typing, launching with ASL-to-English in Gboard and Live Transcribe on the Pixel 11 - turning years of AI research into something immediately useful for a community largely bypassed by the speech-AI boom.
Signing becomes an input method: Users can sign searches, messages and documents, respond in conversations, or interact with Gemini much as hearing users already use voice dictation.
It was trained for scale: Google trained the model on more than 100,000 hours across 50+ sign languages, with further languages and devices planned. That matters because there are more than 200 sign languages used by an estimated 70 million Deaf and hard-of-hearing people globally.
There’s some clever privacy engineering underneath: The phone converts the signer’s movements into geometric body landmarks locally and sends those coordinates for translation, allowing the original camera footage to be discarded rather than uploaded.
Why This Matters: Some of AI’s most consequential applications won’t be about making knowledge workers 7% more productive, they’ll be about removing barriers that technology has quietly imposed on millions of people.
Kismet: Sign language isn’t “English with your hands”: facial expression, head position, torso movement and spatial relationships all carry meaning simultaneously, which means this is genuine language translation wrapped around a pretty formidable computer-vision problem. 👉 Full story here
Electromobility
EVs Are Booming From Denmark to Tunisia - and Petrol Is Starting to Look Lonely
The EV transition is starting to look less like a collection of national experiments and more like a global market reset: 97% of Danish private new-car buyers went fully electric in July, China’s electrified share pushed past 60%, Britain’s used-EV sales jumped 67%, and Tunisia sold more EVs in four months than in all of last year. Different markets, different income levels, same direction, once price, policy and product choice line up, consumers move remarkably quickly.
Northern Europe is approaching endgame: 97% of Danish private buyers chose BEVs in July, alongside Norway at roughly 98%, while Finland and the Netherlands are already approaching half of new-car registrations.
China’s numbers are becoming brutal for combustion: Battery-electric sales rose while pure ICE sales plunged 44% year-on-year, and new-energy vehicles reached roughly two-thirds of retail sales. Chinese EV exports are surging too, which means this transition increasingly doubles as an industrial-policy earthquake.
The mass market is catching up: British used-EV transactions soared 67%, giving ordinary buyers more affordable access, while Tunisia’s sales boom followed sharp tax cuts and cheaper models, a useful reminder that “consumer resistance” often evaporates once the economics make sense.
Why This Matters: EV adoption is no longer advancing in one neat geographic wave; it is spreading through mature markets, used-car markets and emerging economies simultaneously, making electrification look increasingly structural rather than cyclical.
Kismet: Tunisia may be the most revealing datapoint here: change the taxes, cut the sticker price and suddenly a supposedly “unready” market starts buying EVs at twice last year’s annual pace, apparently consumer psychology sometimes has a VAT rate. 👉 Links Inline
The Greenest Car May Not Be the One You Already Own
Here’s one piece of conventional environmental wisdom that may need retiring: new research in Science finds that, from a carbon perspective, replacing a perfectly functional petrol car with an EV sooner is usually better than driving it until it dies. Even swapping out a two-year-old combustion car can cut lifetime emissions by roughly half, because the efficiency gulf between burning petrol and running an electric motor overwhelms the extra emissions from manufacturing the EV surprisingly quickly.
The carbon debt disappears fast: Researchers found it typically takes around three years for the lower running emissions of an EV to compensate for the emissions involved in manufacturing it.
Earlier can actually be greener: Retiring a combustion car in its first year and replacing it with an EV produced the largest modelled benefit, around a 58% emissions reduction over 16 years, even though the researchers deliberately treated the emissions from manufacturing the existing petrol car as already spent.
There are sensible exceptions: Very low-mileage cars, some plug-in hybrids and unusually efficient vehicles running where electricity is especially carbon-intensive may be better kept. But across 92% of the scenarios modelled, replacing a petrol or conventional hybrid vehicle early reduced overall emissions.
Why This Matters: “Use what you already own” remains excellent sustainability advice for almost everything, but combustion cars continually manufacture emissions every kilometre they travel, so delaying electrification can actually lock in more carbon than manufacturing the replacement.
Kismet: A typical combustion engine turns only about 20% of the energy in petrol into movement, with most of the rest disappearing as heat, so keeping an ICE car alive in the name of avoiding waste can amount to carefully preserving a machine whose principal talent is warming the atmosphere and the road around it. 👉 Full story here
Clean Energy
Europe Is Starting to Need Less Gas - Full Stop
For years Europe’s energy-security question was where do we get enough gas? Renewables are starting to rewrite the question entirely: wind and solar are on course to beat gas-fired electricity for the longest stretch ever in 2026, squeezing the traditional gas-burning season and turning “energy security” increasingly into the gloriously boring business of needing less imported fuel in the first place.
The infrastructure has fundamentally changed: Europe has built nearly 750 GW of wind and solar, versus roughly 400 GW of gas generation capacity. Twenty-five years ago, wind and solar barely registered by comparison.
Gas is losing months, not merely market share: Wind and solar increasingly dominate electricity production from roughly April through October, while both coal and gas generation have fallen to multi-year lows. That progressively shortens the period in which utilities need substantial gas-fired generation.
Even Europe’s storage maths could change: Less gas burned for electricity means less gas to import, inject into storage and withdraw later. Storage will remain essential for cold, low-renewables periods for now, but the underlying volume Europe needs as insurance will steadily shrink.
Why This Matters: Renewables aren’t simply cutting power-sector emissions anymore; they’re structurally reducing Europe’s exposure to LNG markets, pipelines, geopolitical shocks and the countries controlling them.
Kismet: Natural gas spent years being sold as the “bridge fuel” between coal and renewables. Europe is now discovering that bridges have supposed another side.
👉 Full story here
India’s Solar Boom Is Eating Coal for Lunch - Literally
India’s extraordinary solar build-out is now visibly pushing coal off the grid: fossil generation at 1pm has fallen by 10 GW in just three years, roughly equivalent to the output of 15 modern coal units. The catch arrives with sunset, evening fossil generation has actually risen, which makes the next stage of India’s transition less about adding solar panels and much more about adding batteries.
Solar is already displacing coal at scale: Midday fossil generation dropped from 135 GW in the first half of 2023 to 125 GW in 2026, despite India’s rapidly rising electricity demand.
The evening peak is now the problem: Between 5pm and 7am, fossil generation averaged 168 GW, up 22 GW in three years, as coal plants ramp to cover demand once solar disappears.
Batteries change that equation: Global storage additions expected this year could theoretically shift 34% of new daily solar generation into evening hours, nearly double 2025’s share and dramatically above the 4% achievable in 2021.
Why This Matters: India is demonstrating that solar can displace coal astonishingly quickly; storage is now the missing piece that determines whether that displacement stops at sunset or continues around the clock.
Kismet: Solar supplied a record 10% of global electricity in the first half of 2026, meaning the technology once dismissed as too intermittent to matter has become the world’s fastest-growing source of electricity, and its biggest remaining weakness increasingly looks like a battery deployment problem rather than a solar problem.
👉 Full story here
Australia Is About to Electrify One of Its Biggest Industrial Monsters
Australia’s largest aluminium smelter, and New South Wales’ single biggest electricity user, is getting a renewable makeover, with Rio Tinto’s Tomago plant set to run entirely on clean electricity by 2033. The price is hardly pocket change: governments are backing it with a A$2.5 billion power-price guarantee, but the deal could unlock nearly 3 GW of new renewables and firming capacity while cutting an extraordinary 7.1 million tonnes of emissions every year.
This is industrial-scale decarbonisation: Tomago consumes more than 10% of all electricity generated in NSW, so switching it from its current coal-heavy supply to renewables moves a genuinely enormous emissions needle.
The power system gets built around it: The agreement is expected to support nearly 3,000 MW of new wind, solar and firming capacity, with batteries helping provide the continuous power an aluminium smelter cannot politely wait around to receive.
There’s a legitimate subsidy debate: Taxpayers are underwriting below-market electricity for ten years to prevent the smelter closing, while its owners have committed A$1.1 billion of investment. Keeping jobs and strategic industry while decarbonising it is attractive; socialising part of the energy bill of a global mining giant deserves scrutiny too.
Why This Matters: Decarbonising electricity generation is one thing; using that clean electricity to keep energy-intensive industries competitive is where the transition starts becoming an industrial strategy rather than merely an energy policy.
Kismet: Tomago alone uses more than a tenth of NSW electricity, so decarbonising a single industrial site will trigger renewable and storage investment at a scale that ripples across an entire state. 👉 Full story here
War - What is it Good For?
Oil’s Supply Problem Is Getting Worse, and Electrification Keeps Looking Smarter
The IEA now expects global oil supply to fall 4.3 million barrels a day this year, with the Strait of Hormuz still effectively constrained, Middle Eastern output sharply below pre-war levels and the third quarter facing the deepest oil-market deficit since 2021. None of this is remotely good news in itself - but every fresh reminder that fossil energy arrives bundled with tankers, chokepoints, wars and price shocks strengthens the economic case for EVs, heat pumps, renewables and electrification.
The shortage is getting deeper: The IEA expects supply to undershoot demand by around 1.27 million barrels per day across 2026, with the July-September deficit reaching roughly 1.8 million barrels per day.
Hormuz remains the enormous vulnerability: Middle Eastern oil loadings briefly recovered towards pre-war levels before falling sharply again, while regional production remained 8.3 million barrels per day below pre-war output in July.
High prices are already destroying demand: The IEA now expects global oil consumption to fall by 1.6 million barrels per day this year, as constrained fuel supplies and higher prices bite. That is the uncomfortable mechanism by which fossil-fuel volatility accelerates substitution: consumers and businesses start looking harder at technologies whose running costs aren’t hostage to the next missile, blockade or tanker attack.
Why This Matters: Clean energy isn’t literally risk-free, but electrification removes an enormous category of fuel-price and geopolitical exposure, and the longer oil markets remain volatile, the stronger the financial case becomes for technologies powered by locally generated electricity rather than globally traded hydrocarbons.
Kismet: The IEA is forecasting falling oil demand at the same time as a severe supply shortage - an unusual combination that shows high prices aren’t merely hurting consumers; they’re actively shrinking the market fossil-fuel producers ultimately depend on. 👉 Full story here
Latest podcasts
Climate Confident:
Soil Is Becoming Infrastructure, and Finance Is Starting to Notice
This week on Climate Confident, I spoke with Holganix’s Tim Weaver about why regenerative agriculture is shifting from a worthy sustainability idea into something CFOs, insurers and investors can actually price. Healthier soil can mean lower fertiliser use, greater water retention, stronger yields and more resilient supply chains, and once those outcomes become measurable, soil starts looking suspiciously like infrastructure.
Water may be the bigger story: Tim argues that increasing soil organic matter can dramatically improve water-holding capacity, reducing drought exposure, runoff and erosion, risks businesses increasingly understand in financial terms.
Measurement is changing the economics: Holganix combines soil cores, probes, farm-equipment data and satellite imagery to measure changes in carbon and organic matter rather than relying purely on models or industry averages.
Farm economics matter more than slogans: Tim says growers in its programmes can cut nitrogen use by 30%, while reduced tillage can lower diesel consumption and healthier soils can support yields, making regenerative agriculture easier to sell when the farmer benefits financially too.
Why This Matters: Regenerative agriculture becomes far more powerful when it stops being framed purely as a carbon intervention and starts being treated as a resilience strategy spanning water, input costs, food security and supply-chain risk.
Kismet: Tim’s most provocative prediction is that soil is having its “Moneyball moment”: once institutional capital becomes comfortable treating measurable soil outcomes as investible assets, regenerative agriculture could move very quickly indeed. 🎧 Listen to the full episode
Resilient Supply Chain:
Supply Chains Don’t Need More Dashboards. They Need Better Decisions.
This week on Resilient Supply Chain, I sat down with Scott DeGroot of the University of Tennessee’s Global Supply Chain Institute and Keith La Londe of PathGuide Technologies to look at the growing gap between seeing what’s happening and actually doing something useful about it. Supply chains have never had more data, yet warehouses still lose time, margin and service because planning systems work on one clock while operations work on another.
“Dashboard theatre” is real: Visibility is useful, but if it doesn’t change what gets picked, packed, staged, loaded or dispatched in the next few hours, it’s mostly expensive decoration.
Near-real-time can beat real-time: Scott’s view has shifted here, organisations don’t necessarily need to ingest everything continuously. They need the right information for the decision window that actually matters, whether that’s the next 12 hours of inbound freight or the next week of labour planning.
AI’s sweet spot is operational grunt work: Labour scheduling, slotting, data cleansing, freight matching and short-horizon optimisation are already ripe for automation, freeing people to focus on exceptions, trade-offs and decisions requiring judgement.
Why This Matters: The next competitive advantage in supply chains won’t come from collecting more data; it will come from shortening the distance between signal, decision and action.
Kismet: One of Keith’s best examples had nothing to do with algorithms: once a warehouse started measuring actual worker productivity, some of the quiet employees nobody particularly noticed turned out to be among its best performers — while one chap who always looked frantically busy wasn’t doing very much at all. 🎧 Listen to the full episode
Don’t forget to follow the podcasts in your podcast app of choice to ensure you don’t miss any episodes.
Featured Chart(s)
Ukraine is helping Russia exceed their Net Zero ambitions, and you love to see it - FAFO in a chart!
EV sales are booming!
EVs require far less energy to move the same distance because they are far more efficient devices.
The energy transition in a chart!
Misc stuff
Given the week that was in it, I thought this road sign saying “Don’t drive with eclipse glasses” was funny!
Given the day that is in it (by-election day in Clacton-on Sea today), I thought this was appropriate
It is a tough choice!
The difference between seeing the movie in 35mm vs IMAX!!!
Obligatory Trump Cartoons
The week a wig went famous and launched 1,000 memes!
Though, if the wig launched 1,000 memes, the Trump hiding in a catering truck while sending his staff and reporters to their possible death on Air Force One launched 100,000 memes!
Engage
If you made it this far, very well done! If you liked this newsletter, or learned something new, feel free to share this newsletter with family and friends. Encourage folks to sign up for it.
Finally, since being impacted by the tech layoffs, I'm currently in the market for a new role. If you know someone who could benefit from my tech savvy, sustainability, and strong social media expertise, I'd be really grateful for a referral.
If you have any comments or suggestions for how I can improve this newsletter, don’t hesitate to let me know. Thanks.
*** Be aware that any typos you find in this newsletter are tests to see who is paying attention! ***
And Finally
We finally discover who has been vandalising Flock cameras?
An entire ad agency in the palm of your hand.
Your next campaign needs a dozen fresh ad variations by Friday. Your agency quotes two weeks and a five-figure invoice. Your in-house designers are already buried under this quarter's requests.
Hightouch Ad Studio fixes that. It reads your brand guidelines, your best-performing creative, and your product catalog, then generates on-brand ads your team can ship the same afternoon. You review and approve every asset before it goes live, so quality holds.
Growth teams use it to build variations for every audience, test more of them, and stop rationing creative because production got expensive. The work that once needed a full agency retainer now runs inside your own workflow, at your pace and under your direction.
You direct the work while Ad Studio handles production, and your designers get their week back.