 |
|
|
|||||
Welcome to AI Decoded, Fast Companys weekly newsletter that breaks down the most important news in the world of AI. You can sign up to receive this newsletter every week via email here. Why Nvidias and AMDs China deal with Trump could backfire The companies making the most money from the AI boom are the ones selling the processors, such as Nvidia and AMD. On Monday those two chip giants cut a deal with the Trump administration that will allow them to sell their products into the China market. For Nvidiathe dominant provider of AI chips powering the generative AI boomthe agreement means that it can once again sell its H20 chip to Chinese developers. Its the latest chapter in a long saga. The Biden administration blocked the sale of Nvidias most powerful AI chips to China in 2022, but deemed the sale of the less powerful H20 chip an acceptable national security risk. The Trump administration continued blocking sales of Nvidias H200 and Blackwell chips. But in April, it went a step further by effectively blocking the sale of the H20 chips, too. The new deal shows that Nvidia CEO Jensen Huangs charm offensive in Washington, D.C., convinced the Trump administration that the U.S.s technological, economic, and national security goals are best served when the worlds AI models and apps are built to run on chips made by U.S.-based companies (like Nvidia). Or maybe the Trump Administration just wanted a piece of the action all along. The administration exploited its jurisdiction over export policy to extract a percentage of Nvidia and AMDs Chinese sales payable to the U.S. Treasury. I said, Listen, I want 20% if Im going to approve this for you, for the country, Trump said during a press conference on August 12, describing his negotiation with Nvidias Huang. He added that Nvidia negotiated the percentage down to 15%. That cut could amount to as much as $3 billion this year, given the high demand for the H20 chips. Trumps demand for 15% is something new in U.S. trade policy. The 15% take indeed sets a precedent, Columbia Business School professor Lori Yue tells Fast Company. It may encourage other companies to adopt similar strategies, viewing profit-sharing as a work-around to government bans. This could lead to a paytoplay arrangement where only the richest corporations can afford to pay the government for permission to sell into a foreign market. Indeed, Treasury Secretary Scott Bessent hinted during a TV interview that the approach could spread to other products in other industries. The administrations tax will very likely translate into higher chip prices for Nvidia and AMDs Chinese customers, such as Tencent. The Chinese government has recently blasted Nvidias H20 chips over security concerns, and the new 15% tax is likely to rankle Beijing even more. The Chinese government would like to see AI developers building their models and apps on top of AI chips from China-based Huawei. The Trump aAdministration may be driving some of Nvidias Chinese customers to do just that, even if it requires rebuilding their infrastructure and tech stacks. Musks Grok chatbot is making noise but falling behind Grok, xAIs chatbot, has had a rough week. The model that powers the Grok chatbot on X is Grok 4, which was announced July 9. At the time, xAI said the new model was the most intelligent one in the world. And the model did achieve state-of-the-art performance on several benchmark tests, including the superhard Humanitys Last Exam, on which it scored 25.4%, three points higher than Googles Gemini 2.5 Pro. As of August 11, the Grok chatbot is available free to all X users, even those on the free tier. In one sense, it marked an immediate improvement. The social platforms permissive environment has made it a clearinghouse for all kinds of misinformation and unsupported claims. Now far more people are using Grok to quickly fact-check those statements. But on August 7 OpenAI released its new GPT-5 model, which outperforms the Grok 4 models on many independent benchmarks. GPT5 shows state-of-the-art reasoning, math and coding skills, visual understanding, creative writing, and health-related question performance. However, GPT5 scored poorly on SimpleBench, ranking 5th behind Grok4 and others in humanlike reasoning and social intelligence. While the benchmark results were circulatingand a GPT-5 backlash was growingxAI CEO Elon Musk threatened to sue Apple for giving OpenAIs ChatGPT app the top ranking among free apps in the App Store (the Grok app ranks fifth). Must presented no evidence, appearing to simply be lashing out. And to make matters worse, Groks problems with inappropriate content resurfaced this week. After being taken offline for spewing antisemitic rhetoric in July, the chatbot was again briefly taken offline Mondaythis time for reasons not divulged by X. The chatbot had an opinion, however, telling one user that it had been taken offline after angrily stating that the U.S. and Israels war on Gaza was a genocide. One user demonstrated that after Grok had been turned on again, it no longer stated that Gaza should be classified as a genocide. Report: Trump Administration stopping tech enforcement During the 2024 presidential campaign, Donald Trump won the support of some tech billionaires, such as Marc Andreessen and Elon Musk, by promising to use a light-touch approach to regulating tech companies, including AI companies. Now Public Citizen is out with a report detailing the extent to which the administration has dialed back oversight and enforcement actions on the tech industry. In just six months with Trump in the Oval Office, the U.S. government has withdrawn or halted 47 of 143 federal enforcement actions against tech companies, the report states. At the start of Trumps second term, 104 technology companies faced 143 federal investigations and enforcement actions. The administration has withdrawn 38 actions and halted 9 others against 45 companies. Beneficiaries include eBay, Meta, Microsoft, PayPal, SpaceX, and Tesla. The report suggests that political spending generated returns through dropped prosecutions and policy changes. Tech corporations, executives, and investors spent $1.2 billion during the 2024 election cycle, including $863 million in donations to super PACs, $222 million in payments to Trump businesses, $76 million in lobbying efforts, and $25 million in inauguration donations, Public Citizen reports. Companies still facing ongoing investigations include Amazon, Apple, ByteDance, Google, Meta, OpenAI, Snap, Uber, Zoom, and various Musk companies (the Boring Co., Neuralink, SpaceX, Tesla, X, and xAI). More AI coverage from Fast Company: Most people are using ChatGPT totally wrongand OpenAIs CEO just proved it AI startup Anthropic is betting on a human editorial team Inside the looming AI-agents war that will redefine the economics of the web AI data wars push Reddit to block the Wayback Machine Want exclusive reporting and trend analysis on technology, business innovation, future of work, and design? Sign up for Fast Company Premium.
Category:
E-Commerce
Imagine you’ve worked hard. You’ve worked long. Mentally, physically you’re done. But then I offer you $1,000 if you can keep going for another 10 minutes. And suddenly you’re good to go. Hold that thought. Mental toughness. Determination. Willpower. Perseverance. However you describe it, the ability to push through mental and physical fatigue to achieve long-term goals is often what sets successful people apart. When others stop trying, the last person to give upespecially on themselvesis often the person who succeeds. But why do some people keep going when others won’t? In part because they understand, if only intuitively, the 40% rule, a concept popularized by Dave Goggins in Jesse Itzlers book Living With a SEAL. The premise is simple. When our minds say we’re exhausted, fried, and totally tapped out, we’re really only 40% done: We still have 60% left in our tanks. So why do we stop? According to a new study published in Journal of Neuroscience, two brain regions activate when you feel mentally fatigued. (Physical fatigue is, in effect, mental fatigue. Your muscles don’t give up when you’re tired. Your mind gives up long before; otherwise you wouldn’t stop until you physically collapse.) One culprit is the right insula, an area deep in the brain associated with feelings of fatigue. The other is the dorsal lateral prefrontal cortex, areas on both sides of the brain that control working memory. All of these areas work together to decide when it’s time to avoid more cognitive effortto decide when you’re done. Except the decisions they make aren’t particularly accurate. According to the researchers: Our study was designed to induce cognitive fatigue and see how people’s choices to exert effort change when they feel fatigue, as well as identify locations in the brain where these decisions are made. However, there may be a discrepancy between perceptions in cognitive fatigue and what the human brain is actually capable of doing. Take incentives. When participants were fatigued, they were more likely to choose to pass up on higher levels of reward for more effort. The more fatigued they became, the greater the reward had to be. But with the right reward, they wouldand obviously couldkeep going. That makes intuitive sense. Deciding to give up is always a benefits/rewards decision. If you’re creating a sales demo, you won’t put in more effort unless you think it’s worth it: If more work will create an appreciably better result, or if the demo’s potential outcome is sufficiently great. Your brain weighs the effort against the outcome. That’s why my offering you $1,000 for 10 more minutes makes you suddenly find mental or physical energy reserves you didn’t think you had. Another factor is time. A study published in PLoS One found that people asked to pedal an exercise bike set at a certain resistance level as long as they could lasted about 12 minutes, until they said they could do no more. But when they were then asked to repeat a five second, maximum-effort power test, they could produce three times more power than they did during the endurance test. Their muscles weren’t depleted. Their mind was depleted. Even if you think youre exhausted, cranking out another five seconds is (relatively) nothing. The endurance test is a different beast. Stuck on a bike, hamster-wheeling away, heart pounding and legs screaming, and not knowing how long the pain will last? Indefinite effort is physically and mentally draining, a combination that makes it much harder to keep pushing past what you perceive as your limit. Even though you could, if offered the right incentive. That’s the primary takeaway. The ability to push through mental or physical fatigue is a trait you can definitely develop. But over the shorter term, if you want to keep going, the key is to find reasonsmeaningful reasonsto stay the course. Getting in better shape so your clothes fit better is a worthwhile goal; getting in better shape so you can live a longer, healthier life and be there for your family is a meaningful goal. Finishing a sales demo because you’ll make the call tomorrow is a worthwhile goal, but crafting a demo that will truly resonate with a potential long-term clientwhich will help you build a thriving businessis a meaningful goal. The key is to find a greater, more long-term “why,” one that will outweigh the feeling that you’re done. And to set time limits on your effort. Not knowing when you’ll be done? You’ll probably decide in, say, 30 minutes, even if you could go longer. But if you say, “I’ll give this one more hour,” now there’s a limit, one you’ll find the energy reserves to reach. When you think you’re done, you really aren’t. Your right insula and dorsal lateral prefrontal cortex have just decided you are. And you can just as easilywith the right incentives, and the right perspectivedecide you’re not done. By Jeff Haden This article originally appeared on Fast Company‘s sister publication, Inc. Inc. is the voice of the American entrepreneur. We inspire, inform, and document the most fascinating people in business: the risk-takers, the innovators, and the ultra-driven go-getters that represent the most dynamic force in the American economy.
Category:
E-Commerce
The NBA on Wednesday unanimously approved the sale of the Boston Celtics to a group led by private equity mogul Bill Chisholm, a deal that values the franchise at $6.1 billion.The league said the transaction is expected to close shortly. When it does, Chisholm will take ownership of at least 51% of the team, with full control coming by 2028 at a price that could bring the total value to $7.3 billion.That’s the largest price ever paid for an American professional sports team, though Los Angeles Dodgers owner Mark Walter has agreed to buy a controlling share of the Lakers in a deal that would place the value of the entire franchise at $10 billion. It’s not clear how much Walter would own; current owner Jeanie Buss is expected to retain at least 15% of the team.The previous record for a U.S. sports franchise was the $6.05 billion paid for the NFL’s Washington Commanders in 2023. The record price for an NBA team was the $4 billion mortgage firm owner Mat Ishbia paid for the Phoenix Suns in 2023.A Massachusetts native and graduate of Dartmouth College and Penn’s Wharton School of business, Chisholm is the managing partner of California-based Symphony Technology Group. The new ownership group also includes Boston businessmen Rob Hale, who is a current Celtics shareholder, and Bruce Beal Jr.Wyc Grousbeck led the ownership group that bought the team in 2002 for $360 million and presided over NBA championships in 2008 and ’24. The franchise’s 18 NBA titles is a record.Chisholm outbid at least two other groups, one led by previous Celtics minority partner Steve Pagliuca. Pagliuca has since announced plans to but the WNBA’s Connecticut Sun for $325 million and move them to Boston, but the women’s league has balked at the deal. AP NBA: https://apnews.com/hub/nba Jimmy Golen, AP Sports Writer
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||