Saturday, August 23rd, 2025

Time	Event
3:30a	Google Says It Dropped the Energy Cost of AI Queries By 33x In One Year Google has released (PDF) a new analysis of its AI's environmental impact, showing that it has cut the energy use of AI text queries by a factor of 33 over the past year. Each prompt now consumes about 0.24 watt-hours -- the equivalent of watching nine seconds of TV. An anonymous reader shares an excerpt from an Ars Technica article: "We estimate the median Gemini Apps text prompt uses 0.24 watt-hours of energy, emits 0.03 grams of carbon dioxide equivalent (gCO2e), and consumes 0.26 milliliters (or about five drops) of water," they conclude. To put that in context, they estimate that the energy use is similar to about nine seconds of TV viewing. The bad news is that the volume of requests is undoubtedly very high. The company has chosen to execute an AI operation with every single search request, a compute demand that simply didn't exist a couple of years ago. So, while the individual impact is small, the cumulative cost is likely to be considerable. The good news? Just a year ago, it would have been far, far worse. Some of this is just down to circumstances. With the boom in solar power in the US and elsewhere, it has gotten easier for Google to arrange for renewable power. As a result, the carbon emissions per unit of energy consumed saw a 1.4x reduction over the past year. But the biggest wins have been on the software side, where different approaches have led to a 33x reduction in energy consumed per prompt. The Google team describes a number of optimizations the company has made that contribute to this. One is an approach termed Mixture-of-Experts, which involves figuring out how to only activate the portion of an AI model needed to handle specific requests, which can drop computational needs by a factor of 10 to 100. They've developed a number of compact versions of their main model, which also reduce the computational load. Data center management also plays a role, as the company can make sure that any active hardware is fully utilized, while allowing the rest to stay in a low-power state. The other thing is that Google designs its own custom AI accelerators, and it architects the software that runs on them, allowing it to optimize both sides of the hardware/software divide to operate well with each other. That's especially critical given that activity on the AI accelerators accounts for over half of the total energy use of a query. Google also has lots of experience running efficient data centers that carries over to the experience with AI. The result of all this is that it estimates that the energy consumption of a typical text query has gone down by 33x in the last year alone. Read more of this story at Slashdot. (Comment on this)
8:34p	Nvidia Release Massive AI-Ready Open European Language Dataset and Tools "Only a tiny fraction of the more than 7,000 languages on Earth are supported by artificial intelligence models," reported SiliconANGLE this week. So Nvidia announced "a massive new AI-ready dataset and models to support the development of high-quality AI translation for European languages." The new dataset, named Granary, is a massive open-source corpus of multilingual audio, including more than a million hours of audio, plus 650,000 hours of speech recognition and 350,000 hours of speech translation. Nvidia's speech AI team collaborated with researchers from Carnegie Mellon University and Fondazione Bruno Kessler to process unlabeled audio and public speech data into information usable for AI training... Granary includes 25 European languages, representing nearly all of the European Union's 24 official languages, plus Russian and Ukrainian. The dataset also contains languages with limited available data, such as Croatian, Estonian and Maltese. This is critically important because providing these underrepresented human-annotated datasets will enable developers to create more inclusive speech technologies for audiences who speak those languages, while using less training data in their AI applications and models... The team demonstrated in their research paper that, compared to other popular datasets, it takes around half as much Granary training data to achieve high accuracy for automatic speech recognition and automatic speech translation. Alongside Granary, Nvidia also released new Canary and Parakeet models to demonstrate what can be created with the dataset... The new Canary is available under a fairly permissive license for commercial and research use, expanding Canary's current languages from four to 25. It offers transcription and translation quality comparable to models three times larger while running inference up to 10 times faster. At 1 billion parameters, it can run completely on-device on most next-gen flagship smartphones for speech translation on the fly. Read more of this story at Slashdot. (Comment on this)
10:34p	Intel's New Funding Came From Already-Awarded Grants. So What Happens Next? The U.S. government's 10% stake in Intel "is a mistake," writes the Washington Post's editorial board, calling Intel "an aging also-ran in critical markets" that "has spent recent years stumbling on execution and missing one strategic opportunity after another." But TechCrunch points out that the U.S. government "does not appear to be committing new funds. Instead, it's simply making good on what Intel described as 'grants previously awarded, but not yet paid, to Intel.'" Specifically, the $8.9 billion is supposed to come from $5.7 billion awarded-but-not-paid to Intel under the Biden administration's CHIPS Act, as well as $3.2 billion also awarded by the Biden administration through the Secure Enclave program. In a post on his social network Truth Social, Trump wrote, "The United States paid nothing for these shares..." Trump has been critical of the CHIPS Act, calling it a "horrible, horrible thing" and calling on House Speaker Mike Johnson to "get rid" of it... According to The New York Times, some bankers and lawyers believe the CHIPS Act may not allow the government to convert its grants to equity, opening this deal to potential legal challenges. Reuters writes that the money "will not be enough for its contract-chipmaking business to flourish, analysts said. Intel still needs external customers for its cutting-edge 14A manufacturing process to go to production, says Summit Insights analyst Kinngai Chan, "to make its foundry arm economically viable." "We don't think any government investment will change the fate of its foundry arm if they cannot secure enough customers..." Reuters has reported that Intel's current 18A process — less advanced than 14A — is facing problems with yield, the measure of how many chips printed are good enough to make available to customers. Large chip factories including TSMC swallow the cost of poor yields during the first iterations of the process when working with customers like Apple. For Intel, which reported net losses for six straight quarters, that's hard to do and still turn a profit. "If the yield is bad then new customers won't use Intel Foundry, so it really won't fix the technical aspect of the company," said Ryuta Makino, analyst at Gabelli Funds, which holds Intel stock. Makino, who believes that Intel can ultimately produce chips at optimal yields, views the deal as a net negative for Intel compared with just receiving the funding under the CHIPS Act as originally promised under the Biden Administration. "This isn't free money," he said. The federal government will not take a seat on Intel's board and has agreed to vote with the company's board on matters that need shareholder approval, Intel said. But this voting agreement comes with "limited exceptions" and the government is getting Intel's shares at a 17.5% discount to their closing price on Friday. The stake will make the U.S. government Intel's biggest shareholder, though neither Trump nor Intel disclosed when the transaction would happen... Some analysts say Intel could benefit from the government's support, including in building out factories. Intel has said it is investing more than $100 billion to expand its U.S. factories and expects to begin high-volume chip production later this year at its Arizona plant. "To have access to capital and a new partial owner that wants to see you succeed are both important," said Peter Tuz, president of Chase Investment Counsel. Read more of this story at Slashdot. (Comment on this)

<< Previous Day

2025/08/23 [Calendar]

Next Day >>