The government’s refusal of industry pleas to raise the AR5 price ceiling was motivated in part by suspicion—exacerbated by glaring informational asymmetries—that their cost complaints were a bluff.

In “de-risking” one set of investment risks—reducing exposure to wholesale market volatility through the creation of regulated auctions and contracts—for for-profit developers, however, the program leaves these same private investment decisions vulnerable to another set of risks: cost.

The new policy removes price guarantees pegged to coal-power prices, with new wind and solar projects supposed to secure direct contracts with electricity buyers. This is likely to lead to lower prices being paid to new wind and solar projects. However, it offers more favourable pricing – via “contracts for difference” – to the amount of new capacity needed to meet central government energy targets.

Climeworks has kept a carbon accounting that the company publishes on its website. It states that it is growing rapidly and now employs 387 people. That is an increase of 45 percent between years. It also says that the company has been expanding systematically and in 2023 it entered new markets, such as the United States, Kenya, Canada, Norway and the United Kingdom. Due to the company's rapid expansion, its carbon footprint has grown in parallel and is attributed to travel and the company's activities.

Central bank should purchase country’s debt if its borrowing costs rise because of excessive risk avoidance by investors

It offers an efficient way

The orphaned well program, which was modeled on a North Dakota initiative, had been widely used by oil states, including several in the West.

Truman had authorized the Atomic Energy Commission (AEC) in 1946 so that the government could identify how nuclear energy might be used, and also reassert some degree of civilian control, now that the war was over. But the AEC would not be fully formed until 1947 and there were still no concrete plans to build a nuclear reactor

This change means many data centers built in central, western, and rural China—where electricity and land are cheaper—are losing their allure to AI companies. In Zhengzhou, a city in Li’s home province of Henan, a newly built data center is even distributing free computing vouchers to local tech firms but still struggles to attract clients.

It is the second biggest producer of electric cars in the world, surpassed only by China. In 2024, the nation built 1.35m electric cars, around a third of total production, according to the German Association of the Automotive Industry.

However, plans to auction 5GW of new gas plants in early 2025 fell apart when the coalition government collapsed. These plants were intended to be “hydrogen-ready”, with the idea they would run on green hydrogen in the future, rather than gas.

The greenhouse gas methane also seeps out of Germany’s lignite mines. Analysis by the thinktank Ember in 2024 concluded that these emissions could be up to 220 times higher than official government reporting suggests.

Lignite is cheap to produce and is used in some of Germany’s coal-power plants. However, it has a relatively low energy content and generates more CO2 when burned than hard coal

Finally, bioenergy – particularly biogas – plays a relatively large role in Germany’s energy system. It has been a key part of the Energiewende, with farmers taking advantage of feed-in tariffs from the 2000s onwards to set up small bioenergy plants.

Germany has also set up a national emissions trading system (ETS)

Around half of the measures relied on government funding, which had dried up following a supreme court ruling that the government’s transfer of €60bn in unused money from a Covid-19 fund to a major climate fund had been unlawful. Another amendment to the climate law came in 2024, pushed by the FDP in the coalition government. This effectively weakened the law by removing sectoral targets, spreading responsibility across the whole government and making it harder to scrutinise progress.

Germany’s target is net-zero for all greenhouse gases by 2045, with interim targets of a 65% emissions cut by 2030 and an 88% cut by 2040, compared to 1990 levels.

The German government announced that it had provided €9.9bn ($10.3bn) of climate finance in 2023. However, Germany has been criticised for relying heavily on loans to deliver climate finance, with the investigative NGO FragDenStaat reporting that €5.1bn ($5.3bn) of the 2023 total was provided in this way.

If there is a silver lining in all this, it’s the casting off of the illusion that private finance would ever prioritize climate goals over profit; in turn, this might mean wider opportunity to refocus on fundamental financial reforms

Over the last 15 years, these scaling laws have held steady, and the compute used to train the most compute-intensive models has reliably increased by roughly 5x each year: over a billion-fold increase.16

2) Globally, the power required by AI data centers could grow by more than 130 GW by 2030, whereas American power generation is forecasted to grow by only 30 GW, much of it unavailable or unusable for AI data centers

And those that will find cause for optimism and relativism, as double materiality is preserved and indeed if 80% of companies are descoped from the CSRD, but in GDP terms, those are 'only' expected to represent 10%.

The readings are always measured values SMETS electricity and gas meters occasionally produce high-values which are not actual recorded consumption. They are error codes used to report some kind of meter fault. You need to trap and handle these when processing the data. As the linked documents note: “Due to the nature of the national programme and the position taken by government and the regulator this is no obligation or requirement for manufacturers to publish these error codes so they can be captured and processed explicitly.“

As Fang Xinghai observed, EU countries “may examine the ESG information disclosed by Chinese firms in levying taxes.” The Carbon Border Adjustment Mechanism (CBAM) – an import tariff based on products’ carbon footprint – is an example of such a tax. Other heated policy debates around “climate clubs” and sustainable sourcing in critical minerals suggest that sustainability-linked trade instruments may become more widespread, at least in Western markets.Such trade-based tools will rely on granular product-level sustainability disclosures, not firm-wide disclosures, which are the subject of the Guidelines and Basic Standards.

I designed a Request for Information (RFI) while at DOE on energy supply chains that was the first to use “online portals” to collect responses in a systematic fashion that could be directly organized into a spreadsheet, rather using an email inbox to collect pdf responses from companies. The portals were really just Microsoft Forms™, but we had to take care not to call them that because saying “form” triggers PRA review. Having our lawyers ensure we did not trigger PRA review delayed the rollout of this by 9 months

the “paperwork reduction act” required any government office to check with the Office of Management and Budget (OMB) in the White House every time they wanted to put out a form, to ensure the form was not duplicative of any other existing form. This measure aimed at “reducing paperwork” from the government, has indeed accomplished its goal, but by making it impossibly slow for any government office to put out any form or survey. This hurts everything from website user surveys to USGS surveys of opinions around public lands.

This might seem obvious now, but at the time it was a bold bet: instead of creating separate product lines optimized for different use-cases (laptops, desktops, IoT, datacenter, etc.), NVIDIA built a single contiguous GPU product line. This meant accepting trade-offs—such as power or cost inefficiencies—but in return, it created a unified ecosystem where every developer’s investment in CUDA could scale seamlessly from gaming GPUs to high-performance datacenter accelerators

On the side of academia France’s research ecosystem boasts notable achievements, such as the 2021 open-source BLOOM model, trained on the publicly funded Jean Zay supercomputer. The model has enabled numerous downstream commercial applications of AI in Europe, such as the Poro-series by the Finnish Silo AI, now owned by AMD.

would require a ramp-up of BECC from currently near-zero commercial capacity to covering almost all biomass conversion.

This flexible power capacity is seldom used, with capacity factors of 28% (Fig. 3), which renders the addition of costly carbon capture to these power plants prohibitively expensive. For carbon capture to be cost effective for a particular technology, high utilization rates are needed due to the high investment cost of the additional infrastructure.

Better Competition: Current competition in the cloud and digital infrastructure market is broken as Europe lacks the equivalent, vertically integrated, cash-generating digital product companies that the US has created by allowing them to fully integrate each part of the value creation chain. Building cloud infrastructure has displaced most other cash generation in these companies and therefore is where they gain their power. By establishing an open market, Europe creates a new, level playing field in which they and EU providers can compete on price and not market access or distribution, which are locked down.

If utilization scales to 65% (more customers are using the AI infrastructure), revenue swells to $795 million (€668 million) per year, while retaining the same staff costs to operate the facilities

Where European providers cannot compete are the following, which are addressed by the market:

Europe should create an open marketplace for buying and selling digital resources, specifically for AI infrastructure. This market should be an independent public, not-for-profit entity financed through transaction fees and with a clear statutory mandate, transparency, and limited leadership terms

In our example, if Siemens scales up its use of the AI provider to €100 million per year, that would mean €50 million worth of digital resources would have to be manufactured at a European facility. That facility pays property taxes on its size, not on the amount of digital resources it produces; therefore, property taxes do not scale with output (that is what profit and trade taxes are for).

These contracts usually fix the price over long periods (15-30 years), shielding them from fluctuations in the energy market at the expense of the power plant. The power plant receives long-term guaranteed income, though it might not benefit from increases in the power price over the next 15-30 years. The investors building the power plants reduce their risk but also forfeit potential value-creation opportunities. Many of them are financed by pension funds.

As far as we can tell from public financial records, even though the request is fulfilled by infrastructure in Europe, the €0.5 million does not travel back through internal transfer pricing between Microsoft's local entities in Europe from the US entity. That means it does not show up as revenue (or profit) in any European entity, although the infrastructure is using European land and power infrastructure.

Beyond this market structure, cloud infrastructure providers are further abusing their position

This is partly due to their ability to purchase large quantities of AI equipment (GPUs) using their vast cash reserves and de facto create a moat by controlling the supply chain for GPUs, hindering incumbents who want to compete.

41.17% of European businesses with more than 250 employees are reported to use AI services. This translates into EU businesses spending an estimated €39 billion a year on AI services

This creates a dependent relationship – in the case of Anthropic, 60-75% of its API revenue comes from the partnership with Amazon AWS.

Carbon dioxide injection companies can now insure against carbon dioxide loss and transfer liability for injection wells to the state after fifty years, which means that the risk of infrastructure failure largely falls on the public

The latter changed almost overnight with the 2022 Inflation Reduction Act (IRA), which almost doubled the tax-credit (called 45Q) for sequestering carbon from $45 per cubic ton of carbon dioxide to $85 (the tax break for DAC is even steeper, at $180). This immediately made many carbon capture projects economically viable, unleashing a scramble for geological pore space suitable for carbon dioxide sequestration

It is no longer possible to claim that capitalists are uninterested; quite the contrary, we are in the midst of a massive speculative CCS boom. I spent the last year studying it wash over Louisiana, which, along with Texas, is at the boom’s epicenter. I attended meetings in communities slated for CCS infrastructure—whether carbon dioxide injection wells, carbon dioxide pipelines or new petrochemical plants that would produce “clean fuels” like blue ammonia and hydrogen using carbon capture.

At the Council on Foreign Relations, Brad Setser and Stephen Paduano have long called for the World Bank to set up an SDR-denominated bond facility, whereby countries like the United States could use their SDRs to help capitalize a massive long-term expansion of the World Bank’s lending headroom

EXIM could also encourage the use of joint venture partnerships to promote responsible technology transfer, both from the United States for technologies like geothermal energy―which is witnessing rapid innovations domestically―and to the United States for technologies like floating offshore wind from countries like Denmark

To that end, the United States has committed to mobilize $200 billion of PGI’s $600 billion goal, and senior White House, State Department, and Treasury officials are helping oversee the program alongside their counterparts across the G7

The Secretary of Defense, in consultation with the Secretary of the Interior, the Secretary of Agriculture, the Secretary of Commerce, and the Secretary of Energy, shall undertake a programmatic environmental review, on a thematic basis, of the environmental effects — and opportunities to mitigate those effects — involved with the construction and operation of AI data centers, as well as of other components of AI infrastructure as the Secretary of Defense deems appropriate.  The review shall conclude, with all appropriate documents published, on the date of the close of the solicitations described in subsection 4(e) of this order, or as soon thereafter as possible

location within geographic areas that are not at risk of persistently failing to attain National Ambient Air Quality Standards, and where the total cancer risk from air pollution is at or below the national average according to the Environmental Protection Agency’s (EPA’s) 2020 AirToxScreen;

possess the characteristics described in subsections (a)(i)-(x) of this section, in a manner that is consistent with the objective of fully permitting and approving work to construct utility-scale power facilities on a timeline that allows for the operation of those facilities by the end of 2027 or as soon as feasible thereafter; and

Within 60 days of the date of this order, the Secretary of Energy shall release for public comment draft reporting requirements for AI data centers covering all phases of AI data centers’ development and operation — including material extraction, component fabrication, transportation, construction, operation, recycling, and retirement — regarding embodied greenhouse gas emissions, water usage, and excess heat or energy expenditures, as distinct from operational intensity of greenhouse gas emissions.

require that, concurrent with operating a frontier AI data center on a Federal site, non-Federal parties constructing, owning, or operating AI infrastructure have procured sufficient new clean power generation resources with capacity value to meet the frontier AI data center’s planned electricity needs, including by providing power that matches the data center’s timing of electricity use on an hourly basis and is deliverable to the data center;

The Secretaries shall, to the extent consistent with applicable law and to the extent that the Secretaries assess that the requirement promotes national defense, national security, or the public interest, as appropriate, select at least one proposal developed and submitted jointly by a consortium of two or more small- or medium-sized organizations — as determined by those organizations’ market capitalization, revenues, or similar characteristics — provided that the Secretaries receive at least one such proposal that meets the appropriate qualifications

In a recent exploration of car bloat, The Economist found that the extra heft of the very heaviest US cars do make their occupants marginally safer, but every life saved corresponds with more than a dozen lost among those inside smaller vehicles that collide with the larger ones. People on foot are at even greater risk.

Browsers can’t just all the sudden make every aspect of a <select> and <option>s styleable, otherwise historical CSS that didn’t apply to them all the sudden does and it would wreak untold havoc on websites. The web doesn’t like to roll like that, and I applaud it for that backwards compatibility. So… there needed to be an opt-in to make it work. A new element can work for that, which for a hot minute seemed like it would be <selectmenu>. But the progressive enhancement story for that basically sucked. So the new opt-in looks like it will be CSS triggered

In the lean-back listening environment that streaming had helped champion, listeners often weren’t even aware of what song or artist they were hearing. As a result, the thinking seemed to be: Why pay full-price royalties if users were only half listening? It was likely from this reasoning that the Perfect Fit Content program was created

it also raises worrying questions for all of us who listen to music. It puts forth an image of a future in which—as streaming services push music further into the background, and normalize anonymous, low-cost playlist filler—the relationship between listener and artist might be severed completely.

Encourages shifting industrial focus to regenerative and circular models, where industries contribute to ecological restoration and local community resilience.

This view supports blended systems where public ownership, cooperative models, and market tools coexist, prioritising affordability and ecological regeneration over profit.

Taking into account central and subcentral taxes, Malta has the highest statutory corporate income tax rate, at 35 percent. Portugal, Germany, and Italy follow, at 31.5 percent, 29.9 percent, and 27.8 percent, respectively. Hungary (9 percent), Ireland (12.5 percent), and Lithuania (15 percent) have the lowest corporate income tax rates.

Membership in the Association of Issuing Bodies (AIB)Issuance of European Energy Certificate System (EECS) Guarantees of OriginGrid connection to another country meeting these criteria

Exhibit A is a new $527 million loan enabling Guyana to take hundreds of local, inefficient oil-fueled generators offline and replace them with an energy efficient combined-cycle gas power plant.

Oil and gas are next for consideration on the no-go list, as reported by our friends over at Grist. That will be much more difficult than cutting off coal, which had a much smaller share of financing to begin with. However, EXIM has already demonstrated its willingness to throw oil under the bus.

Among other projects, ESS is providing its iron-based flow battery to the Nigerian firm Sapele Power, for use at a power plant. The goal is to cut the widespread use of diesel backup generators in Nigeria, by improving efficiency at the power plant.

Growth in data centres and AI servers may diminish the rate of decline but the UK’s high electricity prices do not make the country a logical location for applications that require large amounts of power.

We got off easy here, because logging is a feature of our platform; we run very large ElasticSearch and VictoriaMetrics clusters, fed from Vector and Telegraf, and we’re generally a single ElasticSearch query away from any event happening anywhere in our infrastructure.

We asked peers who had done their own SOC2 and stole their answer: post-facto reviews. We do regular reviews on large components, like the Rust fly-proxy that powers our Anycast network and the Go flyd that drives Fly machines. But smaller projects like our private DNS server, and out-of-process changes like urgent bug fixes, can get merged unreviewed (by a subset of authorized developers). We run a Github bot that flags these PRs automatically and hold a weekly meeting where we review the PRs.

In contrast, regions like LATAM and APAC have tighter guidelines, treating each country as a separate market.

It also flags the potential for “transitional or legacy arrangements” that could protect existing investments from future changes:

The government will therefore aim to secure at least 12GW of new projects over the next two allocation rounds. To enable this, the action plan sets out further reform to the CfD process. 

It also means that the next two “contracts for difference” (CfD) auctions, due to be held in 2025 and 2026, will need to secure the bulk of the offshore wind capacity required for 2030.

Most interestingly, it unlocks ways to avoid coordination within distributed systems, offering better scalability and better performance

DSQL is serverless. Here, we mean that you create a cluster in the AWS console (or API or CLI), and that cluster will include an endpoint. You connect your PostgreSQL client to that endpoint. That’s all you have to do: management, scalability, patching, fault tolerance, durability, etc are all built right in. You never have to worry about infrastructure.

Unlike SQL designs that build on K/V stores, this allows to DSQL to do much of the heavy lifting of filtering and finding data right next to the data itself, on the storage replicas, without sacrificing scalability of storage or compute.

The blog does not detail how the cabinets are connected. Adrian said that, in future, the disaggregated power racks will allow AC inputs to be converted into 400Vdc. Current power solutions convert into 48Vdc, and Adrian argues 400V will be crucial for building more powerful and efficient AI systems. “With 400V we expect improvements and incremental evolution in improved efficiency, like what we have seen in the 48Vdc conversion space,” he said.

The BNetzA gave the example of a consumer using 600 megawatts of electricity per year, which is equal to that of a large and energy-intensive factory. Under the current grid fee scheme, this consumer would have to make a one-off payment of up to 60 million euros. If different levels of grid costs resulting from the connection are factored in, the one-off payment could drop to only twelve million euros if the grid operator assumes an operation cost of, for example, 20 percent of the current standard fee.

The financing strategy was widely used during the 1997-2010 Labour government, and finally ditched by the Conservatives in 2018 after several NHS trusts required bailouts stemming from the high cost of the PFI schemes. Existing contracts were untouched.A report by the National Audit Office, the spending watchdog, that year found taxpayers had incurred billions of pounds in extra costs for no clear benefit through PFIs, with yearly fees running at £10bn.PFI deals are on average 31 years long and generally involve the assets reverting back into public hands at the end. Nearly 71 PFI contracts worth about £4bn are coming to an end in the next four years. More than 300 PFI contracts will expire in the next decadeNumber of PFI contracts expiring, by financial year20502045204020352030202520200604020344 expirations344 expirationsSource: National Audit OfficeA common trigger for clashes when contracts near expiry is the condition of the assets, and the question of who should pay for any deficiencies.

The tech companies, which declined interview requests, said every watt of power they use for their data centers is matched with purchases of clean energy elsewhere on the regional power grid. But those contracts feed into a vast power grid, spanning 14 states from Louisiana to Montana. Many experts and activists say much of that clean power would probably get produced whether the tech companies were signing contracts or not.

In detail, this comprises a change in market dynamics from a “search engine market” to a “web data market”, where web data is used in manifold applications (like social media research, training of AI products, price monitoring, etc.).For this study, we remain aware of such disruptive forcesof altered search, but have not explicitly integrated this in our estimates –although, LLMs still need valid and scalable (web) information for training,which is present in our market model.

If such battery capacity had already been installed this summer, Germany could have displaced 36 GWh of expensive fossil power during evening peaks in June alone.

The role of gas in the evening peak in April 2024 has been roughly halved compared to April 2021.

An interesting heuristic I came up with is that we should install 1 kWh of batteries for every kilowattpeak (kWp) of solar panels. On average solar panels will output around 20% of their rated peak power worldwide (see table). A good rule of thumb is that you want a minimum of 5 hours of storage in the future renewable grid. Five times 20% is 100%, hence 1 kWh of batteries for 1 kWp of solar panels.

Assuming we get all that from renewable energy this would mean an average production of 10 TW (=terawatt). That’s roughly 1 kW per person by the way.

ECL launched its first modular hydrogen data center back in May, with a small deployment at its site in Mountain View, California. Each module supports 1MW, and can cool up to 75kW per rack.For the TerraSite, ECL said that it has three pipelines of hydrogen that will feed the facility, with an energy cost of 0.08-0.12/kWh.

But in 2021, millions of retired carbon credits were tokenized

As the primus inter pares of the Mexican bourgeoisie, Slim has received preferential treatment from the government, which he has reciprocated through constant public appearances with the President. When it became clear that the collapse in 2020 of an elevated metro line that had killed twenty-six people was due to poor workmanship by Mr. Slim’s engineering firm, then-mayor of Mexico City Claudia Sheinbaum reached an agreement with the billionaire: the firm would rebuild the line at a cost of about $40 million dollars. The victims received a compensation of $20,000 to $290,000, depending on whether they had been wounded or killed. For less than $50 million dollars, Slim bought his innocence. 

The movement’s new leader, Claudia Sheinbaum won the Presidency with 60 percent of the votes in early June. With a two-thirds majority in both houses of Congress, the Movement for National Regeneration (Morena) will have the power to completely rewrite the country’s constitutional compact. 

To run a command in the project environment, use uv run. Alternatively the project environment can be activated as normal for a virtual environment.

Significantly, that represents 7.3 TWh of annual clean electricity output that currently does not exist on the US powergrid. Compare that to the annual output of the nation’s largest utility-scale solar plant, Solar Star in California, that generates about 1.6 TWh per year from 747 MW of capacity.

Encouraging more research into sustainable manufacturing and implementing minimum requirements, such as using a certain percentage of renewable energy or investing in eco-friendly chemicals, could be initial steps. This approach mirrors the efforts observed in the implementation of the US CHIPS Act in the United States.

When it comes to measuring the ecological impact of chips during their operation in the end-product, it is difficult to come up with reliable and detailed data. One key challenge is that, according to the GHG Protocol guidance for scope 3, semiconductor manufacturers (both in front- and back-end manufacturing) do not need to report on their climate footprint, particularly with regard to emissions, as their products are classified as ‘intermediate products’ and are not directly sold as end-products

An average hyperscale data centre consumes 11 to 19 million litres of water every day, only half the consumption of front-end manufacturing.

Recently, numerous reports 131 have emphasised the significant electricity consumption in front-end manufacturing by drawing comparisons with homes, 132 cities 133 134 and even national energy usage levels. 135 According to our calculations, the electricity consumption of the European semiconductor industry in 2030 will be around 47.4 tWh (a steep increase from 10tWh in 2021), which is around half that of European data centres (98.5 tWh). 136 Electricity is the biggest single source of GHG emissions

For others, such as hydrogen (H), the picture looks different. The production of H is based on cryogenic processing, which means that the temperature of the gas stream needs to drop to approximately −85°C. 76 This consumes considerable energy and thus emits high indirect GHG emissions in production

Fluorinated gases have a much higher GWP than CO2, and thus a high climate footprint, and account for 80%–90% of direct emissions in a fab. 41 They are reported under scope 1 of the GHG Protocol and are regulated in the European F-gas regulation that was updated in January 2024 and adopted on 7 February 2024. 42

The ecological impact of chemicals stems from three major sources

Fabs in regions with high water scarcity, such as Taiwan, emphasise water reuse and recycling, achieving rates of around 80%, while those in Europe typically have lower recycling rates (10%–14%).

If the goal of a 20% production share stated in the EU Chips Act is met, it is expected that the electricity consumption of the European semiconductor industry will be around 47.4 tWh in 2030, half that of European data centres (98.5 tWh).

But this position is increasingly untenable. If the EU Chips Act achieves its goal of 20% global production by 2030, emissions could increase up to eightfold, surpassing those of other emission-heavy industries. The climate issue will therefore become ever more pressing.

his week I spoke with Matilda Krieder, a researcher at the National Renewable Energy Laboratory, about a database she and her colleagues released this week showing how onshore and offshore wind developers use community benefit agreements

The cost of AI use at any normal human scale doesn’t waste a problematic amount of energy. The problem is the astronomical cost of creating AI systems massive enough to ensure corporate dominance

To GHG Protocol’s credit, the organization does ask companies to report location-based figures alongside their Rec-based figures. Despite that, no company includes both location-based and market-based metrics for all three subcategories of emissions in the bodies of their annual environmental reports.

About a year ago, the Wall Street Journal reported that Microsoft's $10-a-month Github Copilot (which generates code and suggests changes  to the software you're building) loses the company on average $20-a-user-a-month, and in some cases costs Microsoft as much as $80-a-user-a-month. While it's possible that Microsoft could have found ways to make Github Copilot more efficient, this seriously suggests that Microsoft 365 Copilot loses money in much the same way, though generating code is a little more compute-intensive.

Residents rarely learn how data centers may affect their lives until it’s too late. Big tech operators are aggressively deploying nondisclosure agreements to force local officials, construction workers and others to keep these projects under wraps.For tech firms, the incentives to build more of these centers are immense: A McKinsey analysis projected the generative A.I. business could eventually be worth nearly $8 trillion worldwide. Tech companies don’t want to tip off rivals that might try to swoop in and steal a viable site for development. That’s part of the reason they so zealously enforce nondisclosure agreements. But it’s more than that — they also seem to want to avoid angering locals who might learn of the coming disruptions and protest zoning changes.

In August, Germany’s Economy Ministry presented an “options paper” on the future of the country’s electricity market, proposing a mix of liberal and centralised elements of electricity markets, as is the case elsewhere in Europe. Among its proposals, the paper suggests creating a new “capacity market” that would pay generators a flat fee based on size, unlike the current liberal market that rewards them for the energy they sell.

Importantly, receiving nations were required to deposit money equivalent to those products in their central banks. This rebuilt financial health in Europe and order forms for US companies for a generation.

Since then, thousands of people across the country, including military service members, firefighters, and citizens have launched lawsuits against PFAS manufacturers. 3M paid $10.3 billion to settle lawsuits over contaminated drinking water; DuPont paid more than $1.2 billion; Tyco Fire Products, the makers of firefighting foam, settled for $750 million. Altogether, the chemical industry has paid $11 billion to cover the costs of cleaning up ground and water contamination. But so far, no individuals—including the Cotters—have been compensated for the potential effects of exposure to carcinogenic PFAS chemicals. 

Nonetheless, LPG remains an expensive and imported fossil fuel. Indonesia is now drawing up plans for a large-scale deployment of electrified rice cookers and induction stoves.

In short, rough consensuswas an apt description of this informal processin which a proposal must answer to criticisms,but need not be held up if supported by a vastmajority of the group

The IAB increasingly served as thesteward of TCP/IP and the Internet, but had nolegal mandate or enforcement mechanisms. Inother words, IAB-backed protocols were defacto standards, whose status as standardsdepended on broad consensus and widespreadimplementation.

For every euro spent on clean power in Europe over the 2022-40 period, 90 cents of grid investment will be required to achieve the EU’s climate ambitions. 

footrpint

The GLEIF also produces an official mapping file, linking LEI records to the corresponding OpenCorporates records for the same legal entities (and this provides a route from users of OpenCorporates records to BIC and ISIN identifiers, which are also mapped to the LEI). This relationship is underpinned by both organisations having a data model where one-legal-entity is represented by one and only one record; where both organisations have public benefit missions; and where both are committed to non-proprietary identifiers.

Instituting nondiscrimination or common carrier obligations on compute providers operating key points of the stack

The FTC has already outlined this principle in its recent Amazon Alexa case

The UK’s Competition and Markets Authority recently published its initial report on AI Foundation Models, which will be followed up with a subsequent workstream in early 2024

Across the Atlantic, compute power has been an important element in France’s national interest in building out its AI capabilities. Among other moves, France funded the creation of the Jean Zay supercomputer in 2019, operated by the Centre national de la recherche scientifique (CNRS). The computer was used to train BigScience’s BLOOM large-scale AI model

Over 50 new semiconductor projects were announced worth $200 billion following the passage of the Act, according to the Semiconductor Industry Association.162 Among them is TSMC, which plans to make a $40 billion investment in a new facility in Phoenix, Arizona.163 This is particularly notable because it illustrates that market subsidies can function to exacerbate rather than ameliorate market concentration if not carefully calibrated: given the existing bottlenecks in chip fabrication, such investments can easily be captured by dominant players even if they introduce more geographical spread.164 Notably, the chips produced within TSMC’s new facility will still be sent back to Taiwan for packaging and assembly, subverting the hope of creating fully made-in-the-USA chips.165

A 25 percent investment tax credit for semiconductor manufacturing and related equipment

There are a few other pathways toward overcoming the compute barrier, significant among which is a degree of decentralized training. Federated learning could possibly be a way to achieve scale without centralization

For instance, Hugging Face has partnered with AWS in a revenue-sharing agreement to allow developers using Hugging Face to use AWS’s compute and software

Of particular note is that although TSMC is building a chip fabrication facility in Arizona under the US CHIPS Act, all of the chips fabricated at this facility will need to be sent to Taiwan for packaging, meaning that TMSC’s US chip production process will still require a global supply chain network

But this software dominance is also slowly being challenged. OpenAI developed Triton, an open-source software solution that it claims is more efficient than CUDA. Triton can only be used on Nvidia’s GPUs as of now.97 Meta developed PyTorch and then spun off the project as an open-source initiative housed under the Linux Foundation (still financially supported by Meta), and its new version performs relatively well on Nvidia’s A100.98 The benefit of PyTorch is that it can be used across a range of hardware, but on the flip side, it is not optimized for any particular chip

Cerebras differentiates itself by creating a large wafer with logic, memory, and interconnect all on-chip. This leads to a bandwidth that is 10,000 times more than the A100. However, this system costs $2–3 million as compared to $10,000 for the A100, and is only available in a set of 15. Having said that, it is likely that Cerebras is cost efficient for makers of large-scale AI models

Moreover, its proprietary CUDA compiling software is the most well known to AI developers, which further encourages the use of Nvidia hardware as other chips require either more extensive programming or more specialized knowledge.

While FLOP/s grew by more than 6 times between Nvidia’s A100 chips and its latest H100 chips, memory bandwidth (interconnect) only grew by 1.65 times.82 Apart from practical and technological constraints, there is an energy cost to memory bandwidth, with a significant portion of a chip’s energy usage being attributed to interconnect. Overall, this means that interconnect is an important current constraint to the growth of computational power

An important technological constraint with current memory technology is that while logic has only one goal to optimize for (maximizing the number of transistors on a chip), memory is trying to optimize for multiple goals (capacity, bandwidth, and latency).75 Latency has usually lagged behind the other two

This translates directly into cost increases: SOTA AI chips are 10–1,000 times more cost-effective than SOTA CPUs, and 33 times more cost-effective than trailing node AI chips. Thus a large AI model built on trailing node AI chips would be at least 33 times more expensive than models using leading node AI chips.61

in the 1990s and 2000s, the US government reduced its level of investment as interventionist policy fell out of vogue. This set the stage for Intel’s decline relative to firms like Taiwan Semiconductor Manufacturing Company, now the world’s dominant chip fabricator; and ASML, a Dutch company that is the sole manufacturer of the equipment needed to build state-of-the-art chips.

This trend has borne out historically: before the deep learning era, the amount of compute used by AI models doubled in about 21.3 months; since deep learning as a paradigm took hold around 2010, the amount of compute used by models started doubling in only 5.7 months12. Since 2015 however, trends in compute growth have split into two: the amount of compute used in large-scale models has been doubling in roughly 9.9 months, while the amount of compute used in regular-scale models has been doubling in only about 5.7 months

Third, relatedly, we need to excite people with big ideas that are congruent with the crisis, and that simultaneously speak to people’s deep economic and employment anxieties and the cost of living crisis.We need billions of dollars more spent on transformative climate infrastructure that will employ tens of thousands of people.Rather than trying to incentivize heat pumps with inadequate rebates, let’s just make them free! (As PEI does for households with incomes under $100,000.)Let’s talk about free public transit, and huge subsidies for e-bikes, to liberate people from punishing transportation expenses. And let’s propose paying for a chunk of all that with wealth and windfall profits taxes (a recent Abacus survey found increasing taxes on the richest 1% to be a massive vote-winner), and suing the corporations that got us into this mess (as California is doing).These represent transformative policies that tackle multiple crises at once and bolster solidarity.

Who will pick up the $280 billion bill?  So far, it is the U.S. public. As I argued last week, we already effectively own these liabilities. So, how do we get the money to pay for them? 

“It also involves cleaning up more than 100 years of industrial activity and if you look at all these producing fields in the Weald and in the east midlands, it’s a bit like the North Sea.

Utilization effectiveness

Aurora Serverless packs a number of database instances onto a single physical machine (each isolated inside its own virtual machine using AWS’s Nitro Hypervisor). As these databases shrink and grow, resources like CPU and memory are reclaimed from shrinking workloads, pooled in the hypervisor, and given to growing workloads

We weren’t satisfied that only a relatively small number of volumes and customers had better performance. We wanted to bring the benefits of SSDs to everyone. This is an area where scale makes things difficult. We had a large fleet of thousands of storage servers running millions of non-provisioned IOPS customer volumes. Some of those same volumes still exist today. It would be an expensive proposition to throw away all of that hardware and replace it

We didn’t have to worry much about the network getting in the way since end-to-end EBS latency was dominated by HDDs and measured in the 10s of milliseconds. Even our early data center networks were beefy enough to handle our user’s latency and throughput expectations. The addition of 10s of microseconds on the network was a small fraction of overall latency.

These platters have tracks that contain the data. Relative to the size of a track (<100 nanometers), there’s a large arm that swings back and forth to find the right track to read or write your data

Clean Energy Marshall Plan has the makings of a compelling pitch to U.S. domestic audiences: investing in the clean energy transition abroad will benefit businesses and workers at home. Evidence of that effect is already easy to find. The clean investment boom is turning novel technologies into market mainstays: emerging technologies such as hydrogen power and carbon capture now each receive more investment than wind

The Green Climate Fund, the sole multilateral public financial institution devoted to addressing climate change, could follow this approach, too. Almost 15 years after it was founded, the GCF has disbursed only 20 percent of the funding it has received

To complement the Clean Energy Finance Authority, the tariff could be lowered in exchange for foreign procurement of clean energy technologies or of clean products made in the United States

A carbon-based tariff, or a carbon border adjustment, should further motivate climate action by exempting countries that are hitting their nationally determined goals under the 2016 Paris climate agreement or those that fall below certain income and emission thresholds

To accomplish this, the United States must use expanded, stronger, and smarter trade authorities. For example, Washington should build into its tariffs on imported goods an assessment of how much carbon was used to produce them. Tariffs should be determined by the emission intensity of the trading partner’s entire industry, rather than company by company, to avoid “resource reshuffling,” whereby countries try to dodge penalties by limiting their exports to only products manufactured with clean energy instead of reducing their emissions overall. These tariffs should be aimed at all countries, but given its current production practices, China would be hit the hardest.

the term “rare-earth minerals” is a misnomer: these elements are abundant and geographically dispersed. Eighty percent of the world’s lithium reserves, 66 percent of its nickel reserves, and 50 percent of its copper reserves are in democracies. Eighty percent of oil reserves, by contrast, are in OPEC countries, nearly all of which are autocracies

Companies using these services cannot learn by using these digital technologies because they pay only for use, not for access to the intangibles on the cloud.

But the climate impact of data centres could be significantly worse than this. Because of the huge strain data centres are placing on power grids, EirGrid placed a de facto moratorium on new connections around Dublin, causing many to seek a connection to the natural gas network to generate electricity on-site.

The taxonomy is where the rules and data definitions are organised. It is comprisedof a set of elements (i.e., Key Performance Indicatorsand narratives) and all the presentation,calculation and standard logic rules that are in effect. Once created, the XBRL taxonomy is made public as an open sourcefileon the internet. Then, for a specific firm, software can be used to create an XBRL instance (the report itself), containing the specific facts and figures for a certain period. The XBRL instance can be checked against the taxonomy by all parties (reporting entity, a regulator, or even the public) in order toguarantee its data qualityand reliability, as the taxonomy contains data quality checks that any XBRL engine canvalidate

The disclosure required by paragraph 35 shall include the total energy consumption in MWh related to own operations disaggregated by: (a) total energy consumption from fossil sources ( 40 ) ; (b) total energy consumption from nuclear sources; (c) total energy consumption from renewable sources disaggregated by:

DPI can be seen as part of a broader effort to reinvent our relationship to the internet—and, more generally, our digital ecosystem. A large part of its normative appeal stems from the “P” in the acronym: the sense that core functionality on the internet (i.e., identity, payments, data exchange) should not merely serve private ends but rather be reimagined as a set of public goods

This is a good policy, as unbundled REC purchases have a bad reputation when they are used across regions and countries as a cheap substitute for actual investments. But here they are being purchased to match investments in new generating capacity.

To use an extreme and blunt example, if an AI were tasked to stop global warming it might suggest to simply remove all the humans; that might get the job done (solve the task) but not in a way that is aligned with the intent (solve climate change while preserving human life).

In this project, the data workers serve as community researchers. Every community researcher works for two to four months on their inquiry and is compensated for all their working hours. We collaborate with data workers globally. A decisive sampling criterion is that these are data workers who are already organized in workers councils, unions, communities, or advocacy organizations.

On the matter of regulation and legislation, the UK needs to adopt similar legislation to that already in place in the EU namely, the Taxonomy Climate Delegated Act, including the Assessment Framework, the Energy Efficiency Directive and its associated data centre delegated act to collect energy and other environmental data, this could be acheived by simply amending the existing Climate Change Agreement, extending the provisions to all data centres located in the UK and reducing the threshold for compliance reporting to 100kW.

Under the report’s net zero scenario, gas use would peak around the middle of this decade before halving by 2050, compared with 2022 levels. But the current trajectory suggests gas demand will continue to grow throughout the forecast, expanding by about a fifth by 2050.In the scenarios, demand for liquefied natural gas, which is cooled to be transported on ships, climbs by 40% and 30% above 2022 levels respectively.

BP has predicted that the world’s demand for oil will peak next year, bringing an end to rising global carbon emissions by the mid-2020s amid a surge in wind and solar power.

Putting the various figures together shows that, far from the modest 29% year-on-year increase in the incomplete NBS data, there was a record 78% rise in solar generation in May 2024.

Clean energy generated a record-high 44% of China’s electricity in May 2024, pushing coal’s share down to a record low of 53%, despite continued growth in demand. The new analysis for Carbon Brief, based on official figures and other data that only became available last week, reveals the true scale of the drop in coal’s share of the mix. Coal lost seven percentage points compared with May 2023, when it accounted for 60% of generation in China.

They present a significant policy opportunity supported by the country’s ongoing efforts to develop the green bond market and can catalyze promoting green and high-quality development, creating jobs, and delivering environmental benefits.The Greenpeace East Asia report analyzed over 8,000 projects covered by newly issued provincial and municipal government bonds in 2021 as a sample and found that one in five could have been issued as such Green and Sustainable Municipal Bonds.2

“For our customer base, there's a lot of folks who say ‘I don't actually need the newest B100 or B200,’” Erb says. “They don’t need to train the models in four days, they’re okay doing it in two weeks for a quarter of the cost. We actually still have Maxwell-generation GPUs [first released in 2014] that are running in production. That said, we are investing heavily in the next generation.”

As solar is displacing traditional assets, such as gas power plants, we are observing a relatively sharp increase in the price of some of the power reserves, which happens in the opposite direction than the prices on the day-ahead markets.

Germany's coalition government is set to overhaul the way renewable energy is subsidised so that power producers would get one-off support for their investment costs instead of a guaranteed price for power they produce, a finance ministry document showed on Friday.

Mark Boost, CEO of UK-based cloud company Civo is similarly disappointed in the outcome.Boost said the deal is "not good news" for the cloud industry, and added several important questions still need to be answered."We need to know more about how the process of compensation will work," he said. "Will all cloud providers in Europe be compensated, or just CISPE members? Is this a process that will be arbitrated by Microsoft? Where are the regulators in this?"Boost added that the deal will benefit CISPE members only in the short term, but that the cloud industry and its customers will pay the price in the long-term.

However, EFRAG's move to publish a draft for consultation at the same time as the ESRS at the end of January 2024, which is aimed at SMEs not covered by the CSRD, came as a surprise: With the so-called Voluntary ESRS for Non-Listed Small- and Medium-Sized Enterprises the "VSME ESRS", EFRAG is addressing all those companies that are not covered by the CSRD and thus the aforementioned concretisations (i.e., ESRS and ESRS LSME), but which nevertheless wish to take similar measures.

To start, titanium ore is heated to 1,800 degrees Fahrenheit and reacted with chlorine gas and carbon-rich petroleum ​“coke.” This step yields a liquid chemical, titanium tetrachloride, and also produces carbon dioxide as a byproduct (similar to how blast furnaces for ironmaking release CO2).

There also exist well-known vulnerabilities for eBPF programs, which can allow attacksto break container isolation [13] and execute malicious code inthe kernel [22]. Since Wattmeter is built on top of eBPF and ac-cesses RAPL information, only privileged users should be allowedto access it.

Figure 3 shows the effect of EFS on the CPU and energy sharesbetween the two processes

Another framework for implementing custom scheduling poli-cies without direct kernel modification is Meta’ssched_ext.sched_extis a Linux kernel patch that proposes a new abstract schedulerclass, which can be instantiated with scheduling programs in eBPFat runtime. There are ongoing discussion of upstreamingsched_extinto Linux. We have chosen to implement the scheduling poli-cies proposed in this paper with ghOSt, but it is also possible toimplement them withsched_ext

He added that private companies, which will have a significant role in the transition, want to see policy certainty enhanced in the months ahead. The group is awaiting the finalisation of the Electricity Regulation Amendment Bill, which promises to open up the electricity market and put an end to Eskom’s longstanding monopoly, and the Integrated Resource Plan.

consumes 19% less energy per event in high performance mode

%. This shows, that the impact of load shaping heavily depends on the power proportionality53 of the underlying hardware, and that it is not a reasonable measure per se.

Although solar forecasts are not very promising this time, it again permits to discharge to 30%, as the carbon intensity during the next day is expected to be especially low. Instead of drawing carbon-intensive grid energy at night, the demand is thereby shifted to the next morning where the batteries are charged to 60% using cleaner energy.

Although the carbon-aware experiment uses 2.4 % more energy than the baseline (which is because not all power modes have the same energy-efficiency), its associated carbon emissions through grid power consumption are 35.9 % lower. In the following, we will briefly analyze the two experiments and to demonstrate how our integration enables research and development of carbon-aware applications.

The control unit adaptively adjusts the battery's minimum state of charge and grid charge rate over time. In particular, in case of promising forecasts for solar power production or low carbon intensity, it is able to temporarily deplete the battery to 30%.

This is especially the case when not testing virtualized applications on powerful hardware but embedded systems that often only run only one energy-hungry process at a time. In these systems, load shaping is likely rather performed on a device level, for example, through DVFS.

Therefore, if applications under test are deployed on physical nodes like single-board computers, it is recommended to use dedicated hardware for measuring the power usage of these devices. For example, in our experimental testbed we monitor the current and voltage of a RaspberryPi 4b with a USB to USB measuring device equipped with an INA219 DC current sensor.

Note: The physical node's power usage is controlled via DVFS, while the virtual node uses rate limiting on the executed process.

“These claims assume that a company that pollutes more now should be able to pollute into the future. This means Global North companies will continue to inequitably dominate use of our remaining carbon budget. These findings should raise real questions for any bodies that claim to set standards for voluntary corporate climate targets,” David Tong, Global Industry Campaign Manager for Oil Change International, said.

A major blindspot is the fact that SBTi does not take into account new companies with their share of emissions coming into existence in the future. In SBTi’s framework, existing companies are allocated a share of the carbon budget without leaving any room for new players, some of whom might be more efficient or even working in the decarbonisation space like solar technologies. This further entrenches fossil fuel developments by existing companies and also raises questions about equity.

Net-zero corporate pledges are voluntary, which means they can be reeled back in as quickly as they are announced.Shell scrapped its emission reduction target for 2035 when it sought to grow its gas business, for example, and BP walked back on some of its climate commitments when profits hit a record high. An Oil Change International assessment of the climate plans of eight oil majors—Chevron, ExxonMobil, Shell, TotalEnergies, BP, Eni, Equinor, and ConocoPhillips—released this year found that all eight continue to drive fossil fuel expansion and six have explicit goals to grow their total production volume this decade

Not all do so and there is no agreement on what a fair share is, but by claiming an individual target is Paris compliant is implicitly making an ethical claim that this represents your fair share of the global response, without saying how it is fair,” she said.

The Climate Policy paper points to issues that arise when individual countries or companies link their efforts to mitigate climate change to Paris Agreement goals. If such linking is indeed necessary, the authors say that assumptions about time scale, spatial scale and equity must be included in the analysis and presented transparently.The paper is “a welcome contribution” because it asks countries to center mitigation claims in a context of national contributions to global fair shares of the climate response, said Kate Dooley, a research fellow in the School of Geography, Earth and Atmospheric Sciences at the University of Melbourne. At present, many wealthy countries have exceeded their fair share of the carbon budget.

The distribution of emissions over time is not just a question of cumulative global emissions—it’s also a matter of equity. Rich, industrialized countries have already claimed a disproportionate share of the carbon budget that has brought us to around 1.2°C warming today. The U.S. has emitted about 25 percent of cumulative emissions; Europe, meanwhile, has emitted around 20 percent.

In fact, capturing CO2 from ethanol production is so easy that ethanol production is the biggest source of purified CO2 in the US, and the second biggest source in Europe. This is where we source the CO2 that we use in fizzy drinks, food production, fire extinguishers, etc.

In fact, capturing CO2 from ethanol production is so easy that ethanol production is the biggest source of purified CO2 in the US, and the second biggest source in Europe. This is where we source the CO2 that we use in fizzy drinks, food production, fire extinguishers, etc.