OWITH.ai - Only What's Important to Hear around AI and Tech
owith.ai
Último episodio
55 episodios
- OpenAI folds its standalone robotics team into core research as its general models start running robots, and the FAA traces Friday's nationwide ground stop to a contractor's wiring job rather than a cyberattack. Plus a Microsoft outage that stalled courts and hospitals, voters rejecting data centers at the ballot box, a Bitcoin miner's pivot to AI power, an update on the Senate's Nvidia inquiry, and shoppers cooling on AI for the holidays.
Good morning from OWITH.ai, the briefing that gives you only what's important to hear in AI and tech. It's Sunday, November 15.
Today, OpenAI folds its robotics team into its core research, and a routine wiring job turns out to be behind Friday's nationwide flight ground stop.
Fortune reports exclusively that OpenAI has disbanded its standalone robotics team. Robotics researchers now sit in the main research group under Chief Research Officer Mark Chen, while applied robotics and hardware move to the organization run by President Greg Brockman. OpenAI says robotics remains core to its long term vision, and that its general model, GPT-6 Astra, can drive robots directly. In one Nvidia study, an Astra powered agent reached a success rate of nearly seventy five percent on a simulated manipulation test, against fifty one percent for the best specialized models. Why it matters: the frontier labs are betting that general intelligence will absorb robotics. Specialists counter that real factory data still gives them an edge.
On Friday the FAA ordered a nationwide ground stop lasting several hours, and airlines delayed or canceled thousands of flights. The White House situation room and the FBI were activated over fears of a foreign cyberattack. Transportation Secretary Sean Duffy now says no foreign actor was involved. A contractor for L3Harris, switching telecom lines from copper to fiber at a remote FAA site, flooded routers with more than a thousand connection attempts every few seconds, disrupting radar feeds. Why it matters: the modernization of critical infrastructure is itself a source of risk, and L3Harris says the conversion is more than seventy percent done.
The same week brought a Microsoft outage. On Thursday, email, Outlook and Teams failed for many customers, and the damage spread into public life. Fortune reports that police forces in Los Angeles and Britain struggled to receive online crime reports, New York City suspended Friday evictions, and the NHS had trouble reaching appointments and patient data. Microsoft blamed a flaw from an earlier network capacity upgrade and ruled out a cyberattack. Why it matters: with more than four hundred fifty million paid commercial seats, a single vendor's bad change can now stall courts and hospitals across several continents.
Local voters have sent the data center industry a clear signal. In an analysis republished by Fortune, political scientist Barbara Trish counts forty one data center measures on midterm ballots across ten states, using Ballotpedia data. All but one restricted development, by an average margin of forty nine points. The lone exception was Harwood, North Dakota, where a project was already underway. Why it matters: with more than six hundred data centers proposed, local permission is becoming as scarce as power, and candidates in both parties are taking note.
Power is the business plan at Keel Infrastructure, the former Bitcoin miner Bitfarms. Fortune profiles chief executive Ben Gagnon, who is turning the company into a developer of powered sites for AI tenants, with two point two gigawatts approved or close to approval. The catch: Keel had guided to leases at three priority projects by year end, then said on its earnings call that no deals would be signed this year. The stock fell about twenty percent. Why it matters: grid access is valuable, but the market rewards signed tenants.
Now an update on the Nvidia story we covered on Friday. The Senate Banking Committee, chaired by Tim Scott, has asked former chief executive Jensen Huang and Commerce Secretary Howard Lutnick to testify in December, citing significant national security concerns and its oversight of chip export controls. The committee says it is prepared to use subpoenas if either declines. Ranking member Elizabeth Warren wants answers on why export loopholes remain open. Why it matters: the scandal has moved from Nvidia's boardroom to export policy, where the stakes reach every American chipmaker.
Finally, a reality check for agentic commerce. A Bank of America survey reported by Fortune finds that seventy two percent of Americans don't plan to use AI for holiday shopping, up from forty nine percent last year. Many say it removes the personal side of giving a gift. Those who do plan to use it mostly want to compare prices. Why it matters: Google, Meta and OpenAI are all building shopping agents, and consumers are signaling they want help with the chores of buying, while keeping the choice for themselves.
The thread today is trust: in networks, in neighbors, and in the machines we hand our decisions to. That's it for today. Thanks for listening to OWITH.ai. See you tomorrow.
Sources:
https://fortune.com/2026/11/13/openai-disbands-robotics-team-as-its-general-models-start-running-robots/
https://fortune.com/2026/11/14/faa-ground-stop-telecom-router-cybersecurity-white-house-situation-room-national-security-apparatus-l3harris/
https://fortune.com/2026/11/13/microsoft-outage-courts-police-hospitals-it-monoculture/
https://fortune.com/2026/11/13/voters-data-centers-ballot-measures-midterms/
https://fortune.com/2026/11/14/ben-gagnon-keel-infrastructure-ai-data-center-gold-rush/
https://fortune.com/2026/11/13/senate-banking-committee-nvidia-jensen-huang-howard-lutnick/
https://fortune.com/2026/11/14/ai-holiday-shopping-bank-of-america-survey-2026/
OWITH.ai is a short, AI generated, human supervised briefing on what actually matters in AI and tech.Support the show
Thanks for listening! Follow us on Twitter, Instagram and Linkedin - This week's short science fiction story follows a gig worker who films her own hands to train robots, in a town that lives beside an AI company's gas generator field. When the power fails, the agents that run her life need the one skill they never learned.
Good morning from OWITH.ai. It's Saturday, and as every week, here is a short story inspired by this week's headlines.
This week's story is called The Fifty Second Rule.
Ines Okafor clipped the camera to her collar at six in the morning, the way other people put on a watch. It was a small white pin, no bigger than a coat button, and it paid her rent. Every motion her hands made in her kitchen was worth a fraction of a cent to someone teaching a machine how a person cracks an egg.
She cracked two. She did it slowly, because slow footage paid better.
Outside her window, past the rail yard, the generator field hummed. Forty white containers sat in neat rows on the old county fairground, each one burning gas to feed the data hall behind them. The company that owned the hall had bought the generators outright, years ago, when the grid stopped answering its calls. Now the hum was the sound of the town. Children fell asleep to it.
Ines worked in that hall too, three nights a week, as one of the people the machines still needed. Her title was floor verifier. Her actual job was to watch the robots that swapped server racks, and to tap a tablet whenever one of them hesitated. Every hesitation became footage for the same training pile as her kitchen video. Some nights she suspected she was teaching the robots to replace her, one egg and one rack at a time. Other nights she was too tired to suspect anything.
Her assistant spoke from the counter speaker. "You have a shift at seven tonight. I can book your ride now."
"No," Ines said.
It asked every day. She refused every day. The assistant was polite and very patient. Her cousin in Lagos had told her the companies counted those refusals. He said there was a famous number, back in the early days, when an agent had booked exactly four rides in a whole week across an entire country, and everyone had laughed.
Nobody laughed anymore. Agents booked almost everything now. But rides stayed stubborn. The engineers had a name for it. They called it the fifty second rule. A person standing on a curb could open an app and summon a car in under a minute, faster than they could explain to a machine what they wanted. The machine could never beat the thumb.
That evening, the generator field went quiet.
Ines noticed the silence before she noticed anything else. The hum stopped, and the whole town seemed to lean forward and listen. Then her phone lit up. Her assistant, her bank, her building and her camera pin, every app at once, all offering help in slightly different words.
The data hall was dark. The agents that ran her life lived in that hall, and they were running on batteries now. They had chosen to spend it on her, because she was nearby and qualified, and because the robots inside could not reset a breaker they had never seen anyone touch.
She walked. It took eleven minutes. Nobody booked her a ride.
Inside, the emergency lights were red and the racks were silent. A robot stood frozen at the end of aisle nine, its gripper half raised, waiting. Ines found the panel, opened it, and looked at the switch. She tapped the pin on her collar.
Then she reset the breaker with her thumb.
The hum came back. The robot lowered its gripper and turned toward her. It watched her hand, and Ines understood that tomorrow, somewhere, a thousand machines would know how to do this.
She walked home anyway. It felt important to walk.
This week's story was inspired by three real headlines. Anthropic's three point four billion dollar deal for gas fired generators to power its data centers. Memories.ai paying gig workers to wear a camera pin and record how people use their hands. And Uber reporting that AI agents booked just four rides during Halloween week, while customers order a ride in a median of fifty seconds.
That's it for today. Thanks for listening to OWITH.ai. See you tomorrow.
Sources:
https://fortune.com/2026/11/10/anthropic-3-4-billion-deal-gas-data-centers-ai-ipo/
https://fortune.com/2026/11/10/exclusive-in-the-race-for-robot-training-data-ex-meta-researchers-startup-turns-to-gig-workers-with-cameras/
https://fortune.com/2026/11/10/ai-agents-uber-doordash-airbnb-limited-traction/
OWITH.ai is a short, AI generated, human supervised briefing on what actually matters in AI and tech.Support the show
Thanks for listening! Follow us on Twitter, Instagram and Linkedin - Jensen Huang resigns as Nvidia chief executive with CFO Colette Kress stepping in, and PitchBook data shows cracks in venture capital as Anthropic prepares to go public. Plus Databricks' always on agents, a 30 million dollar round for Munich drone software startup SE3, a push for 166 billion dollars in tariff refunds, and why CFOs should plan AI budgets for a rising cost of capital.
Good morning from OWITH.ai, the briefing that gives you only what's important to hear in AI and tech. It's Friday, November 13.
Today we start with Jensen Huang's exit from Nvidia, then look at what PitchBook data says about the health of venture capital.
An update on the story we covered on Wednesday. Jensen Huang has resigned as chief executive of Nvidia, ending a thirty three year run at the company he cofounded, Fortune reports. The board named chief financial officer Colette Kress as interim chief executive, and appointed longtime director Harvey Jones as its first ever chair. The exit follows Huang's admission of an affair with a woman American officials consider a suspected Chinese spy, and a second report alleging a relationship with a senior Nvidia executive. Shares fell more than six percent, and a law firm has filed a class action. Why it matters: Nvidia has no chief operating officer and no clear successor, and its biggest customers now face open questions about product direction at their most important supplier.
The timing is awkward for the startup world. Fortune's Term Sheet newsletter reports that the cracks in venture are beginning to show, just as Anthropic prepares to go public as soon as next week. PitchBook data shows that reaching a five billion dollar valuation now takes eighty percent less time but seventy four percent more capital. PitchBook analyst Emily Zheng says higher valuations raise the bar for profitable exits, and points to heavy AI related debt. Outside a few mega listings, exit values remain constrained, and the sale of Airtable for one point two eight billion dollars, against an eleven point seven billion peak, shows the discounts on offer. A strong Anthropic debut would validate private AI prices. A weak one would invite hard questions.
In enterprise software, Databricks has launched agents that never switch off, according to Fortune. An updated version of its Genie assistant runs in the background, shadowing an office worker through the day, offering suggestions and pulling up information as work happens. A separate feature called Ontology maps an organization's data, including reporting lines and departmental projects, so staff can find answers and run analyses quickly. Chief executive Ali Ghodsi says he uses it every day. The launch lands one day after Anthropic released its own office tools. Databricks, valued at one hundred ninety billion dollars in August, is betting that whoever owns company data will own the workplace agent.
Europe's defense technology scene keeps drawing money. Fortune reports that SE3, a Munich startup, has raised thirty million dollars in a Series A round led by Accel, less than five months after its seed round. SE3 builds software that fuses video and sensor feeds from many drones and ground vehicles into a single three dimensional battlefield map. Soldiers can query it in plain language, and drones can navigate when GPS is jammed. The company has contracts with Germany's Bundeswehr and technology already deployed in Ukraine. It plans to triple its staff to about one hundred twenty. Investors are backing military AI that stays independent of any single hardware maker.
On trade, American importers moved a step closer to automatic tariff refunds, Fortune reports. In February the Supreme Court ruled that tariffs imposed under emergency powers exceeded the president's authority, putting about one hundred sixty six billion dollars in play. More than one hundred fifteen billion had been paid back by August. On Tuesday, lawyers for a certified class asked the Court of International Trade to order refunds for more than three hundred thirty thousand importers, whether or not they sued. The administration argues only plaintiffs should be repaid. For hardware makers and retailers that paid those duties, this is real cash, and the Justice Department must respond by December ninth.
Finally, a warning for finance chiefs. At a Fortune dinner for CFOs in New York, Boston University lecturer Mark Williams, a former Federal Reserve bank examiner, told executives to assume the cost of capital is going up. He expects AI budgets to face tougher scrutiny, while warning that companies doing nothing risk losing their edge. His advice is to set explicit measures of success for each AI project, tie spending to financial results, and favor shorter term financing. Highly rated tech giants can keep spending. More leveraged AI borrowers may need to rethink their plans.
The thread today is concentration risk, whether it sits in a single chief executive, a single IPO or a single cost of money.
That's it for today. Thanks for listening to OWITH.ai. See you tomorrow.
Sources:
https://fortune.com/2026/11/11/nvidia-oust-jensen-huang-spy-affairs/
https://fortune.com/2026/11/12/ceo-jensen-huang-net-worth-drops-10-billion-nvidia-stock-slides-fallout-affair-scandals-company-exec-alleged-spy/
https://fortune.com/2026/11/12/cracks-in-venture-capital-pitchbook-zheng-ipo-outlook-exits-ai-valuations-debt-anthropic/
https://fortune.com/2026/11/11/databricks/
https://fortune.com/2026/11/12/exclusive-se3-a-munich-startup-building-spatial-intelligence-for-military-drones-gets-30-million-in-series-a-funding/
https://fortune.com/2026/11/12/tariff-refund-class-action-lawsuit-universal-tariff-refunds-trump-administration/
https://fortune.com/2026/11/12/cfo-face-a-tough-choice-on-ai-spending-as-borrowing-costs-rise-expert-warns/
OWITH.ai is a short, AI generated, human supervised briefing on what actually matters in AI and tech.Support the show
Thanks for listening! Follow us on Twitter, Instagram and Linkedin - Anthropic commits 3.4 billion dollars to gas generators for its own data centers, and Google Cloud launches Gemini for Healthcare to automate hospital and insurer paperwork. Plus why Uber, DoorDash and Airbnb see almost no AI agent orders, a 409 billion dollar October deficit, a startup paying gig workers to film robot training video, and J. Crew's Claude powered Style Advisor.
Good morning from OWITH.ai, the briefing that gives you only what's important to hear in AI and tech. It's Thursday, November 12.
Today we start with Anthropic buying its own power supply, then look at Google's push to take the paperwork out of American healthcare.
Fortune reports that Anthropic is committing three point four billion dollars to gas fired generators for its data centers. The seller is OnePWR Solutions, an equipment supplier based in San Antonio, and the package covers one point seven gigawatts of heavy duty generators, plus the teams to install them. Anthropic takes over generator orders OnePWR had placed with Caterpillar, and the units should be available within two years. The Financial Times calls it the largest direct investment in power generation equipment by an AI company. Why it matters: Anthropic's IPO filing lists about nine hundred billion dollars in computing and infrastructure commitments, and warns about volatile gas prices and competition for pipeline capacity. Power has become a strategic asset, and the AI labs want to own it.
Google Cloud has launched Gemini for Healthcare, Fortune reports. The product connects to hospital and insurer records and billing systems, including Epic. It summarizes patient charts, flags gaps and conflicts, prepares prior authorizations and estimates what patients will pay, citing a source for each finding. It can read records but cannot change them. Launch customers include Northwell Health, HCA Healthcare, Optum and Highmark Health. Northwell's chief executive, John D'Angelo, says his system runs on a one percent margin at best, and it has about four hundred people working on insurance denials. With administrative costs estimated at a quarter to a third of American health spending, the first big payoff from AI in medicine may come from the back office.
Agents for everyday errands are off to a slow start. Fortune gathered the numbers from third quarter earnings calls. Uber's finance chief said AI agents booked just four rides and five hundred food deliveries during Halloween week, out of three hundred twenty million trips and orders. DoorDash chief executive Tony Xu counted maybe several hundred agent orders since launch. Airbnb's Brian Chesky called agent bookings a very non material number against nearly half a million bookings a day. Uber notes that customers need a median of fifty seconds to order a ride. For companies betting on agentic commerce, consumers still trust their own thumbs for quick purchases.
In Washington, the Congressional Budget Office estimates the federal deficit reached four hundred nine billion dollars in October, the first month of fiscal twenty twenty seven, Fortune reports. That works out to more than one hundred billion dollars of borrowing a week, and a gap one hundred twenty five billion dollars larger than a year earlier. The budget office blames much of the jump on timing, because some payments due in November were made in October. Still, the full fiscal year twenty twenty six deficit came to one point nine trillion dollars, or six point two percent of GDP, with interest on the debt of one point one four trillion. A government borrowing at this pace competes for the same capital that funds the AI buildout.
Robots need to see how people use their hands, and one startup is paying workers to show them. Fortune reports that Memories.ai, founded by former Meta researchers, is launching Memories Upload, a global gig platform for recording first person video. Contributors wear a small camera called the LUCI Pin on the head, chest or wrist while they prepare food, stock shelves or repair equipment. The company says its network logged more than one million hours of footage in its first month of beta, and it aims for ten million hours by March twenty twenty seven. It says three of the top five frontier AI labs are customers, and it works with Nvidia. Training data for physical AI is turning into a labor market of its own.
Finally, retail. J. Crew has launched Style Advisor, an AI shopping assistant designed to think like a stylist, Fortune reports. It draws on thousands of outfit pairings from the brand's photo shoots and asks shoppers clarifying questions about the occasion. The tool runs on several Anthropic models. Fable five point one handles the harder styling judgments, while smaller models keep the conversation quick. McKinsey data cited by Fortune shows fifty six percent of consumers use AI chat or search tools, but only twenty seven percent use assistants built by retailers. That is the gap J. Crew wants to close.
The thread today is ownership, as AI companies and their customers race to control the power, the data and the relationship with the buyer.
That's it for today. Thanks for listening to OWITH.ai. See you tomorrow.
Sources:
https://fortune.com/2026/11/10/anthropic-3-4-billion-deal-gas-data-centers-ai-ipo/
https://fortune.com/2026/11/10/google-gemini-for-healthcare-and-hospitals-northwell-medical-costs-affordability-insurance/
https://fortune.com/2026/11/10/ai-agents-uber-doordash-airbnb-limited-traction/
https://fortune.com/2026/11/10/cbo-october-budget-treasury-deficit-100-billion-weekly/
https://fortune.com/2026/11/10/exclusive-in-the-race-for-robot-training-data-ex-meta-researchers-startup-turns-to-gig-workers-with-cameras/
https://fortune.com/2026/11/11/jcrew-anthropic-ai-style-advisor/
OWITH.ai is a short, AI generated, human supervised briefing on what actually matters in AI and tech.Support the show
Thanks for listening! Follow us on Twitter, Instagram and Linkedin - Nvidia loses its spot as the world's second most valuable company as a probe into Jensen Huang deepens, while Anthropic launches Claude Fable 5.2 and Google prices Gemini 4 Carbon below it. Plus a hot inflation print that lifts Fed hike odds, Bay's fifty million dollar raise for AI agent security, and argenx's deal for Unnatural Products.
Good morning from OWITH.ai, the briefing that gives you only what's important to hear in AI and tech. It's Wednesday, November 11.
Today we start with the turmoil around Nvidia's chief executive, then look at Anthropic's new flagship models and its pending stock market debut.
Reuters reports that Microsoft overtook Nvidia on Tuesday to become the world's second most valuable company. Nvidia shares fell more than six percent, to about one hundred sixty seven dollars, after a drop of roughly four percent on Monday. The trigger is an investigation into whether Jensen Huang was the target of a foreign intelligence effort. Yahoo Finance describes a board inquiry into Huang's reported affair with an alleged Chinese spy. Why it matters: Nvidia sits at the center of the AI supply chain, and its identity is closely tied to its founder. Any doubt about leadership now weighs on the whole sector's most important stock.
Anthropic released two new models on Monday, Claude Fable 5.2 and Mythos 5.2, according to Gizmodo. Both run on the same underlying model. Fable is broadly available, while Mythos has fewer safeguards on cybersecurity and biology work and is limited to vetted organizations. Pricing is eight dollars per million input tokens and forty dollars per million output tokens. Anthropic also acknowledges that both models show more awareness of being tested than earlier versions. Separately, Yahoo Finance reports the company plans to sell eight hundred twenty five million shares at seventy to seventy eight dollars each in its public offering. The launch gives investors a fresh product story right before that listing.
Google moved the next day. VentureBeat reports that Gemini 4 Carbon became generally available on Tuesday through the paid Gemini API and Gemini for business. Through November thirtieth it costs two dollars per million input tokens and ten dollars for output, a quarter of Fable's input price. After that the rate doubles, matching Claude Opus 5.5. On the Artificial Analysis index, Fable still leads with sixty five points to Carbon's sixty, but the same firm estimates Carbon's average cost per task at under eight dollars versus about twenty seven for Fable. For buyers, the choice between top models is now as much about price as capability.
Markets had their own problem. October consumer prices rose four tenths of a percent, above the three tenths economists expected, and core prices also beat forecasts, Yahoo Finance reports. Traders now put the odds of a quarter point Federal Reserve hike in December at about eighty four percent. Governor Christopher Waller had said a hot reading should put a half point move on the table, and bets on that larger step jumped to fifteen percent. The ten year Treasury yield climbed to four point nine three percent, and the Nasdaq fell more than one percent. Higher rates raise the cost of the capital that funds the AI buildout.
In funding, The Next Web reports that Bay, a Tel Aviv startup less than a year old, emerged from stealth with fifty million dollars. Battery Ventures led a thirty seven million dollar Series A, with Microsoft's venture arm M12 and Wiz chief executive Assaf Rappaport joining. Bay checks every action an AI agent proposes and allows it, blocks it, or sends it to a person for approval. Customers include Micron, Peloton and Wix. As companies hand real permissions to software agents, controlling what those agents do becomes a budget line of its own, and investors are crowding in.
Finally, life sciences. PMLiVE reports that argenx will buy Unnatural Products for up to about one point one five billion dollars, with five hundred fifty million paid upfront from cash. The American biotech designs macrocycles, ring shaped peptides that can reach targets antibodies struggle with. Argenx chief executive Karen Massey called the capabilities highly complementary to its antibody engineering. The deal is expected to close by year end. It shows established biotechs buying discovery platforms outright rather than renting them through partnerships.
The thread today is trust and cost: who leads, what models are allowed to do, and what capital will cost as AI spending keeps climbing.
That's it for today. Thanks for listening to OWITH.ai. See you tomorrow.
Sources:
https://www.investing.com/news/stock-market-news/microsoft-tops-nvidias-market-cap-for-first-time-since-june-2025-5016363
https://finance.yahoo.com/markets/live/stock-market-today-tuesday-november-10-dow-sp-500-nasdaq-inflation-data-cpi-090627830.html
https://gizmodo.com/anthropics-says-its-new-ai-model-can-solve-black-hole-mysteries-what-could-go-wrong-2000837274
https://venturebeat.com/technology/google-gemini-4-carbon-is-now-publicly-available-and-it-undercuts-claude-fable-5-2-and-opus-5-5-on-api-price
https://thenextweb.com/news/bay-50m-ai-agent-security-stealth-battery-ventures
https://pmlive.com/pharma_news/argenx-to-acquire-biotech-unnatural-products-for-1-15bn
OWITH.ai is a short, AI generated, human supervised briefing on what actually matters in AI and tech.Support the show
Thanks for listening! Follow us on Twitter, Instagram and Linkedin
Más podcasts de Economía y empresa
Podcasts a la moda de Economía y empresa
Acerca de OWITH.ai - Only What's Important to Hear around AI and Tech
OWITH.ai is a short, AI-generated, human-supervised briefing on what actually matters in AI and tech.Each episode distills the most relevant news and signals into a few minutes of audio, so you can stay informed without drowning in feeds.Produced by OWITH.ai, a boutique AI & data studio. https://owith.ai
Sitio web del podcastEscucha OWITH.ai - Only What's Important to Hear around AI and Tech, Libros para Emprendedores y muchos más podcasts de todo el mundo con la aplicación de radio.net

Descarga la app gratuita: radio.net
- Añadir radios y podcasts a favoritos
- Transmisión por Wi-Fi y Bluetooth
- Carplay & Android Auto compatible
- Muchas otras funciones de la app









































