The Dutch government's 2027 budget for its Ministry of Economic Affairs and Climate, published on 15 September, says the cabinet is making a one-off 120 million euros available from 2027 for companies taking part in the Germany-led IPCEI-AI, a European artificial intelligence project under the Important Projects of Common European Interest scheme. The budget says the money follows a survey of company interest and that the project builds on an earlier one for cloud services and infrastructure. For people in the Netherlands nothing changes yet; this is money set aside in a budget, not yet spent.
Brazil signs its data centre tax law into force, twelve days after the Senate passed it 66 to nil
· Band Rules and society · Lei Complementar 237, signed with one article vetoed · Confirmed
Brazil's president signed Lei Complementar 237 into law on 15 September with a partial veto, enacting the data centre tax measure the Senate approved unanimously, 66 votes to nil, on 3 September. The law amends Lei Complementar 229 so that a special taxation regime for data centre services sits among the exceptions to the rules restricting how governments grant tax benefits, alongside Brazil's fiscal responsibility law. It grants no specific tax break itself. What it does is make it structurally easier for individual Brazilian states and cities to offer one to data centre operators without hitting the fiscal limits an ordinary incentive would face, so long as it still fits the annual budget. Article 3 was vetoed.
Cognition, which builds the AI coding agent Devin, and Amazon Web Services announced a multi-year strategic collaboration on 15 September to expand how large companies buy and deploy Devin through AWS. No deal size is given. The companies offer customer results instead: at Mercedes-Benz they say Devin analysed more than 200,000 lines of COBOL, reducing an estimated eight-month modernisation project to eight days, and at an unnamed global automaker it moved a 25,000-line COBOL workflow off a mainframe to AWS Lambda at an estimated 73 percent lower cost. Both are labelled estimates by the companies themselves. COBOL is an old language still running inside banks and carmakers, and translating it has turned out to be one of the clearer uses for coding agents so far, more so than writing new code.
Factory, which builds AI agents for software engineering teams, said on 15 September that it has raised 200 million dollars at a 5 billion dollar valuation, taking total funding past 400 million. The money comes from Blackstone, Khosla Ventures, Sequoia Capital, Insight Partners, Evantic Capital, Sound Ventures, NEA, Mantis VC and Clearlake. The company names Nvidia, Blackstone, Royal Bank of Canada, Palo Alto Networks, Adobe and T-Mobile as customers, which means Blackstone appears on both sides of this round, as investor and as user. Factory says it is used by hundreds of thousands of developers. Its release states no prior valuation, so how far that 5 billion has moved is not something this desk can check. Valuations across AI coding-agent companies have run up all year as software teams hand routine work to agents and investors bet on which one becomes the default.
A quarterly survey of 705 software developers across more than 60 countries and 41 enterprise chief technology officers, run by the IT services company BairesDev, found 42 percent of developers now say AI writes at least half their code, up from 12 percent a year earlier. Developers report saving 13 hours a week on coding, up from about 7. The counterweight is in the same survey: 67 percent say they now spend more time reviewing AI-generated code, and only 7 percent say the decision to ship has been handed to AI without their input. So the work is moving rather than disappearing, from writing to checking. This is a company-commissioned survey rather than an official statistic, and BairesDev sells software engineering services, but the size of the year-on-year shift is a useful gauge of how fast AI-written code is spreading through ordinary teams.
A new poll finds just 27 percent of Americans approve of the president's handling of AI
· Band Rules and society · 11% strongly and 16% somewhat approve; 44% strongly and 11% somewhat disapprove · Confirmed
A YouGov poll of 3,564 US adults, conducted and published on 15 September, asked how people rate President Trump's handling of artificial intelligence. YouGov publishes the four components rather than totals: 11 percent strongly approve, 16 percent somewhat approve, 11 percent somewhat disapprove, 44 percent strongly disapprove, and 18 percent are not sure. Adding those gives 27 percent approval against 55 percent disapproval, and the shape matters as much as the total: strong disapproval alone is four times strong approval. This measures views of the administration's handling of AI as a policy area, not of AI itself or of any company. Public opinion on AI policy is one of the pressures shaping how fast and how hard lawmakers move on the industry, in statehouses and in Congress.
Google's new slow-thinking voice model takes the top spot on an independent speech-quality index
· L4 Models · 82.6 on the Speech-to-Speech Quality Index, ranked #1; 97.7% on Big Bench Audio · Confirmed
Google released two voice models on 15 September: Gemini 3.8 Live, and a slower variant called Extended Thinking that reasons before it answers. The scores belong to the slower one. Google reports Extended Thinking at 82.6 on Artificial Analysis's Speech-to-Speech Quality Index, which it says is the top overall spot, plus 97.7 percent on Big Bench Audio and 68.6 percent on a voice agent benchmark called tau-Voice. Plain Gemini 3.8 Live took second place in the Speech Agent Arena, which ranks by user preference. Two caveats worth carrying: these are Google reporting its own placement on outside indexes rather than those evaluators publishing a ranking, and Google describes the price only as competitive against other frontier models, without saying whether it has moved from its own previous generation. Outside indexes are still one of the few ways to compare voice systems across companies, since labs mostly publish their own favourable tests.
China's industry ministry and its top economic planning agency jointly published a five-year plan on 15 September 2026 that sets a goal of more than 30 trillion yuan, about 4.47 trillion dollars, in revenue at the country's larger electronics firms by 2030. The plan organises 17 tasks under four headings, and the first task under strengthening the industry's foundations is what the document calls full-chain breakthroughs in integrated circuits, meaning China wants to design and make chips at every stage rather than assemble them from imported parts. It also sets a target of 3.5 percent of revenue going to research. For people outside China, it is Beijing's stated plan for the industry that makes the world's electronics.
Micron demonstrates the world's first 512 gigabyte memory module for AI servers
· L2 Silicon · 512GB DDR5, 12TB per server, 16.0W vs 44.2W · Confirmed
Micron says it has demonstrated the world's first 512 gigabyte memory module of a type called DDR5, meant for the servers that run AI systems. With these, a single two-processor server could hold up to 12 terabytes of memory, and Micron measures one module at 16 watts against 44.2 watts for the four smaller modules it replaces, more than 60 percent less. AMD and Intel are still validating it and volume production is planned for the second half of 2027, so this is a preview of the next generation of AI servers rather than something you can buy. More memory per server, drawing less power, is one of the few levers the industry has against data centre electricity bills rising even faster than they already are.
Profound, a startup that helps companies track and try to influence how they are mentioned when people ask AI chatbots for recommendations, raised 180 million dollars in a Series D at a 1.8 billion dollar valuation. That is less than seven months after its 96 million dollar Series C, which independent reporting valued at about 1 billion dollars, so the valuation has grown roughly 80 percent in that time, our own arithmetic from the two figures. Profound says it now serves over 1,000 enterprise brands and more than a third of the Fortune 100. If you research products online, this is part of why the answers AI chatbots give about brands are increasingly managed and optimised rather than neutral, the way search results already are.
AIUC, short for Artificial Intelligence Underwriting Company, raised 40 million dollars in a Series A led by Ribbit Capital with participation from First Harmonic, which takes its total funding to 55 million dollars alongside an earlier 15 million dollar seed. One founder was Anthropic's first product hire; the other was a partner in McKinsey's insurance practice and chief operating officer of the AI safety research group METR. The company has built a standard called AIUC-1 that tests how agents hold up against risks including jailbreaks, hallucinations and data leaks, using 5,000 risk and attack combinations. Its own release names Cursor, ElevenLabs, Harvey, KPMG, Lovable, UiPath and Fin as companies that certify against it. For most people nothing changes today; this is an attempt to build the kind of inspection standard insurers and large companies want before they rely on AI agents more heavily.
Euclyd, a Dutch startup building specialised chips for running already trained AI models rather than training new ones, signed a Series A financing round of more than 200 million euros, co-led by Samsung, Somerset Capital Partners and Scaleup Europe Fund. Peter Wennink, the former president and chief executive of ASML, the Dutch company that makes the machines used to manufacture advanced chips, joins as chairman. It is the second large round this month for a company betting that specialised, cheaper chips beat general purpose ones at running trained models, after Positron AI's 875 million dollars on 10 September, our own observation rather than a comparison Euclyd makes. For most people nothing changes yet; this money funds chip development.
India's government says 12 semiconductor projects have been approved across six states, with investment commitments of more than 1.64 lakh crore rupees, roughly 18 billion dollars on this desk's conversion at about 91 rupees to the dollar, and that three facilities have already started commercial production. The projects span silicon and compound semiconductor fabs, display manufacturing and advanced packaging. This is a different figure from the 1,27,500 crore rupee Semicon 2.0 push in this site's Pulse for 17 September; the government describes that as the outlay it approved in July to scale this foundation, while this counts projects already approved. The figures came in a government backgrounder two days before Prime Minister Modi opened SEMICON India 2026. For ordinary Indians, most of these are approvals rather than working factories so far; if they are built, the effect over the next several years is more chips made in India and more jobs making them.