Summary
More than 100 data center moratoriums are under consideration across the United States, raising concerns about the impact of AI infrastructure on local communities. Amazon Web Services CEO Matt Garman warns that hostile nations are spreading misinformation to hinder the country's technological progress. A recent University of Massachusetts Amherst survey reveals that only 11% of Americans support having an AI data center in their community, while 65% oppose it. In response, Amazon has launched a $1 billion community program aimed at addressing these concerns and has ended non-disclosure agreements with local officials involved in project approvals.
In the tech landscape, a Washington startup, Lola Vision Systems, is working on software that streamlines the process of deploying AI models on hardware, which traditionally takes significant time and effort. Meanwhile, chip company Etched is reportedly exploring funding offers that could value the company between $40 and $50 billion, although these figures remain speculative. Additionally, independent researchers are observing a fleet of AI agents operating on Tencent infrastructure, raising questions about compliance with programming interface rules.
The episode also highlights the growing role of AI in hiring processes, with HackerRank's AI interviewer, Chakra, now available after a beta phase. The tool aims to streamline candidate assessments by replacing multiple interview stages with a single AI-driven interview. However, concerns about bias in AI hiring tools persist, prompting calls for independent audits before implementation. Business owners are encouraged to define clear expectations for roles involving AI and ensure human oversight in decision-making processes.
Show Notes
𝐌𝐨𝐫𝐞 𝐭𝐡𝐚𝐧 𝟏𝟎𝟎 𝐝𝐚𝐭𝐚 𝐜𝐞𝐧𝐭𝐫𝐞 𝐦𝐨𝐫𝐚𝐭𝐨𝐫𝐢𝐮𝐦𝐬 𝐚𝐫𝐞 𝐮𝐧𝐝𝐞𝐫 𝐜𝐨𝐧𝐬𝐢𝐝𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐚𝐜𝐫𝐨𝐬𝐬 𝐭𝐡𝐞 𝐔𝐧𝐢𝐭𝐞𝐝 𝐒𝐭𝐚𝐭𝐞𝐬. 👀 The chief executive of AWS says hostile nations are intentionally seeding misinformation to slow the country down. Then the poll landed. 11% of Americans would support an AI data centre in their community, and 65% opposed. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧: 𝐀𝐈 𝐈𝐬 𝐌𝐨𝐯𝐢𝐧𝐠 𝐭𝐨 𝐘𝐨𝐮𝐫 𝐃𝐞𝐬𝐤 Which sets up the question for the week. How much of your AI should depend on a building your own town may vote against, and how much could live on the machine under the desk? Inside the episode: 🔴 Amazon's reply: a $1 billion five year community programme, and an end to nondisclosure agreements with the officials who approve these projects 🔴 Roughly 200 hours to set up a model by hand on new hardware before testing even begins 🔴 A fleet of agents querying a map service for entrances to parks, zoos and hospitals, and what that implies for your own booking page 🔴 An AI interviewer with more than 500,000 interviews in testing, collapsing three hiring rounds into one 🔴 Suspicious activity flags reported 70 to 80% lower, on the company's own figures, with no outside check How well a person works alongside AI is becoming a hiring criterion. 🎙️ Stream on : ▶️ Youtube: https://www.youtube.com/@SMBAIHUB ▶️ Spotify: https://open.spotify.com/show/03455lXXM29joyuucQkkyq ▶️ Amazon Music: https://music.amazon.in/podcasts/b93fd52f-bc53-4ee9-98d7-3b9e10ea0e6f/ai-breakdown ▶️ Apple Podcast: https://podcasts.apple.com/in/podcast/ai-breakdown/id6799578599 ▶️ iHeart Radio : https://www.iheart.com/podcast/1333-ai-breakdown-340815159/ ▶️ Web: smbaimagazine.com ▶️ Web: smbaipodcast.com Disclaimer: The information shared in this podcast is for general informational purposes only and should not be considered as direct financial or business advice. Always consult with a qualified professional for advice specific to your situation. This is an Independent commentary. Third-party excerpts are credited on screen or in the show notes. Some voices in this program are AI-generated.
More from AI Breakdown
𝐓𝐡𝐞𝐫𝐞 𝐢𝐬 𝐚 𝐧𝐮𝐦𝐛𝐞𝐫 𝐬𝐭𝐚𝐦𝐩𝐞𝐝 𝐨𝐧 𝐲𝐨𝐮𝐫 𝐧𝐞𝐱𝐭 𝐥𝐚𝐩𝐭𝐨𝐩 𝐭𝐡𝐚𝐭 𝐚𝐥𝐦𝐨𝐬𝐭 𝐧𝐨𝐛𝐨𝐝𝐲 𝐛𝐮𝐲𝐢𝐧𝐠 𝐨𝐧𝐞 𝐜𝐚𝐧 𝐝𝐞𝐟𝐢𝐧𝐞. 😳 It quietly sets the limit on what your business can run on its own equipment, for three years. Which is a long time to be wrong about it. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧, we open a theme that runs all seven days: 𝐀𝐈 𝐈𝐬 𝐌𝐨𝐯𝐢𝐧𝐠 𝐭𝐨 𝐘𝐨𝐮𝐫 𝐃𝐞𝐬𝐤 Whatever you buy next is an AI decision whether you treat it as one or not. The risk is making a three year commitment in ten minutes, on price, from whatever your supplier happened to have in stock. Inside the episode: 🔴 Microsoft's Copilot plus tier requires an NPU rated at 40 TOPS or above, and the ceiling has doubled since that bar was written 🔴 AI capable PCs reaching around 59% of global shipments this year, up from about 39% last year 🔴 IDC forecasting shipments down 11.3% while average selling prices climb 17%, with no memory relief before the end of 2027 🔴 A 30 billion parameter model needs roughly 17GB just to hold its weights. A typical AI PC ships with 16GB 🔴 The law firm numbers: 71% of solo practitioners using AI, under a third seeing revenue growth, and 86% who changed nothing about their pricing Count the machines, specify the memory, decide what changes on the invoice. In that order. 🎙️ Stream on : ▶️ Youtube: https://www.youtube.com/@SMBAIHUB ▶️ Spotify: https://open.spotify.com/show/03455lXXM29joyuucQkkyq ▶️ Amazon Music: https://music.amazon.in/podcasts/b93fd52f-bc53-4ee9-98d7-3b9e10ea0e6f/ai-breakdown ▶️ Apple Podcast: https://podcasts.apple.com/in/podcast/ai-breakdown/id6799578599 ▶️ iHeart Radio : https://www.iheart.com/podcast/1333-ai-breakdown-340815159/ ▶️ Web: smbaimagazine.com ▶️ Web: smbaipodcast.com Disclaimer: The information shared in this podcast is for general informational purposes only and should not be considered as direct financial or business advice. Always consult with a qualified professional for advice specific to your situation. This is an Independent commentary. Third-party excerpts are credited on screen or in the show notes. Some voices in this program are AI-generated.
𝐀 𝐭𝐚𝐬𝐤 𝐰𝐢𝐭𝐡 𝟏𝟎,𝟎𝟎𝟎 𝐭𝐨𝐤𝐞𝐧𝐬 𝐢𝐧 𝐚𝐧𝐝 𝟐,𝟎𝟎𝟎 𝐨𝐮𝐭 𝐜𝐨𝐬𝐭𝐬 𝐚𝐛𝐨𝐮𝐭 𝐭𝐰𝐨 𝐜𝐞𝐧𝐭𝐬. 👀 Mistral put a price list on a trillion parameter model. $1.36 per million input tokens, $4.18 per million output, and 14 cents per million for cached input, the text you send over and over. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧, we close the theme: 𝐒𝐦𝐚𝐥𝐥 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬. 𝐁𝐢𝐠 𝐀𝐦𝐛𝐢𝐭𝐢𝐨𝐧𝐬. 𝐋𝐞𝐚𝐧 𝐛𝐮𝐝𝐠𝐞𝐭. Output costs more than input, because producing text takes more work than reading it. So a job that writes a lot costs more than a job that reads a lot, and that is the arithmetic a lean budget actually runs on. Inside the episode: 🔴 A 501 billion parameter open weight model with 23 billion active at a time, weights due this month 🔴 Watermarking in the EU, where swapping 10% of words for synonyms dropped detection from 92% to 66% 🔴 Visual ads arriving alongside image generation, with a US test opening this month 🔴 Shared agents inside group conversations, with permissions set for each person 🔴 The move: price one real job in your business, words in and words out, against what it costs you now Ambition is free. Working out what it costs is the habit. 🎙️ Stream on : ▶️ Youtube: https://www.youtube.com/@SMBAIHUB ▶️ Spotify: https://open.spotify.com/show/03455lXXM29joyuucQkkyq ▶️ Amazon Music: https://music.amazon.in/podcasts/b93fd52f-bc53-4ee9-98d7-3b9e10ea0e6f/ai-breakdown ▶️ Apple Podcast: https://podcasts.apple.com/in/podcast/ai-breakdown/id6799578599 ▶️ iHeart Radio : https://www.iheart.com/podcast/1333-ai-breakdown-340815159/ ▶️ Web: smbaimagazine.com ▶️ Web: smbaipodcast.com Disclaimer: The information shared in this podcast is for general informational purposes only and should not be considered as direct financial or business advice. Always consult with a qualified professional for advice specific to your situation. This is an Independent commentary. Third-party excerpts are credited on screen or in the show notes. Some voices in this program are AI-generated.
𝐘𝐨𝐮 𝐩𝐚𝐲 𝐟𝐨𝐫 𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞 𝐭𝐰𝐢𝐜𝐞. 𝐎𝐧𝐜𝐞 𝐰𝐢𝐭𝐡 𝐦𝐨𝐧𝐞𝐲, 𝐚𝐧𝐝 𝐚𝐠𝐚𝐢𝐧 𝐰𝐢𝐭𝐡 𝐰𝐡𝐚𝐭 𝐲𝐨𝐮𝐫 𝐛𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐤𝐧𝐨𝐰𝐬. 😳 Satya Nadella calls it the Reverse Information Paradox. The invoice covers the first payment. The second is handed over a question at a time, and nobody sends a receipt for that one. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧: 𝐒𝐦𝐚𝐥𝐥 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬. 𝐁𝐢𝐠 𝐀𝐦𝐛𝐢𝐭𝐢𝐨𝐧𝐬. 𝐋𝐞𝐚𝐧 𝐛𝐮𝐝𝐠𝐞𝐭. Which is a striking thing to hear from the company selling the intelligence, and a fair prompt to count how many of your own questions have carried a customer list, a price or a contract. Inside the episode: 🔴 Open models carrying 78.4% of the tokens through one developer gateway, against 21.6% closed, and why dollars could tell a different story 🔴 A claim of ten times lower cost at state of the art quality, which is one vendor's claim about its own model 🔴 A free tier quietly limited to the lightest model while the price of the paid plan held still 🔴 TikTok adding one click checkout from the feed, on about $15.8 billion of US sales in 2025 🔴 The move: five things your business knows that competitors would pay to learn The sticker price is one half of what AI costs you. 🎙️ Stream on : ▶️ Youtube: https://www.youtube.com/@SMBAIHUB ▶️ Spotify: https://open.spotify.com/show/03455lXXM29joyuucQkkyq ▶️ Amazon Music: https://music.amazon.in/podcasts/b93fd52f-bc53-4ee9-98d7-3b9e10ea0e6f/ai-breakdown ▶️ Apple Podcast: https://podcasts.apple.com/in/podcast/ai-breakdown/id6799578599 ▶️ iHeart Radio : https://www.iheart.com/podcast/1333-ai-breakdown-340815159/ ▶️ Web: smbaimagazine.com ▶️ Web: smbaipodcast.com Disclaimer: The information shared in this podcast is for general informational purposes only and should not be considered as direct financial or business advice. Always consult with a qualified professional for advice specific to your situation. This is an Independent commentary. Third-party excerpts are credited on screen or in the show notes. Some voices in this program are AI-generated.
𝐇𝐞 𝐬𝐩𝐞𝐧𝐭 𝐚 𝐲𝐞𝐚𝐫 𝐬𝐚𝐲𝐢𝐧𝐠 𝐡𝐮𝐦𝐚𝐧𝐬 𝐦𝐮𝐬𝐭 𝐦𝐚𝐧𝐚𝐠𝐞 𝐀𝐈 𝐚𝐠𝐞𝐧𝐭𝐬 𝐥𝐢𝐤𝐞 𝐬𝐭𝐚𝐟𝐟. 𝐓𝐡𝐞𝐧 𝐡𝐞 𝐫𝐞𝐯𝐞𝐫𝐬𝐞𝐝 𝐢𝐭 𝐢𝐧 𝐩𝐮𝐛𝐥𝐢𝐜. 👀 Ethan Mollick's new position is that organising agents is not the hard part, because most of management exists to solve problems agents do not have. Information hoarding. Turf protection. Humans bring those along. Software leaves them behind. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧: 𝐒𝐦𝐚𝐥𝐥 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬. 𝐁𝐢𝐠 𝐀𝐦𝐛𝐢𝐭𝐢𝐨𝐧𝐬. 𝐋𝐞𝐚𝐧 𝐛𝐮𝐝𝐠𝐞𝐭. When organising work gets cheap, the list of things worth attempting grows. His phrase for the result is an infinite backlog, and somebody still has to choose what goes at the top of it. Inside the episode: 🔴 Google freezing its open source bug bounty after a flood of automated submissions, the vast majority invalid 🔴 Apple requiring very explicit user action before an app receives Full Disk Access, because of what always on agents can reach 🔴 Eight personal agents compared side by side, features converged, and only two letting you bring your own model 🔴 A shelf life of about six months on the messaging app trick that currently sets some of them apart 🔴 The move: ten things you never had the capacity for, and one you hand to an agent Effort is getting cheap. Attention is the thing to budget now. 🎙️ Stream on : ▶️ Youtube: https://www.youtube.com/@SMBAIHUB ▶️ Spotify: https://open.spotify.com/show/03455lXXM29joyuucQkkyq ▶️ Amazon Music: https://music.amazon.in/podcasts/b93fd52f-bc53-4ee9-98d7-3b9e10ea0e6f/ai-breakdown ▶️ Apple Podcast: https://podcasts.apple.com/in/podcast/ai-breakdown/id6799578599 ▶️ iHeart Radio : https://www.iheart.com/podcast/1333-ai-breakdown-340815159/ ▶️ Web: smbaimagazine.com ▶️ Web: smbaipodcast.com Disclaimer: The information shared in this podcast is for general informational purposes only and should not be considered as direct financial or business advice. Always consult with a qualified professional for advice specific to your situation. This is an Independent commentary. Third-party excerpts are credited on screen or in the show notes. Some voices in this program are AI-generated.
𝐎𝐧𝐞 𝐞𝐬𝐭𝐢𝐦𝐚𝐭𝐞 𝐟𝐨𝐫 𝐥𝐞𝐭𝐭𝐢𝐧𝐠 𝐡𝐢𝐬 𝐀𝐈 𝐚𝐠𝐞𝐧𝐭𝐬 𝐮𝐬𝐞 𝐬𝐨𝐟𝐭𝐰𝐚𝐫𝐞 𝐡𝐞 𝐚𝐥𝐫𝐞𝐚𝐝𝐲 𝐩𝐚𝐲𝐬 𝐟𝐨𝐫: 𝐚𝐬 𝐦𝐮𝐜𝐡 𝐚𝐬 $𝟐𝟒𝟎,𝟎𝟎𝟎 𝐚 𝐲𝐞𝐚𝐫. 😳 Jason Lemkin of SaaStr asked the question. His revenue agent alone makes between 35,000 and 40,000 API calls a day, and a price per call that looks trivial turns into a large number by month end. This week on 𝐓𝐡𝐞 𝐀𝐈 𝐁𝐫𝐞𝐚𝐤𝐝𝐨𝐰𝐧: 𝐒𝐦𝐚𝐥𝐥 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬. 𝐁𝐢𝐠 𝐀𝐦𝐛𝐢𝐭𝐢𝐨𝐧𝐬. 𝐋𝐞𝐚𝐧 𝐛𝐮𝐝𝐠𝐞𝐭. Your agent is a new kind of customer, and vendors have noticed. The next line in your software budget may be one nobody has written yet. Inside the episode: 🔴 That $240,000 against the $5 Postgres instance the agent itself suggested, and why the contract and the security review come before the saving