OpenAI’s New Silicon, Oracle’s 21,000 Job Cuts, and Getty’s 145% Stock Surge

Yesterday in AI  |  June 25, 2026

OpenAI’s New Silicon, Oracle’s 21,000 Job Cuts, and Getty’s 145% Stock Surge

The infrastructure behind frontier AI is changing forever as tech giants move toward vertical integration. Today's episode breaks down OpenAI's surprise announcement of its first custom inference chip, Jalapeño, built with Broadcom to challenge Nvidia's dominance and slash scaling costs ahead of its upcoming IPO.
 
Plus, we unpack the historic corporate disclosure inside Oracle's SEC filing, where the tech giant openly attributed 21,000 eliminated jobs directly to AI deployment. We explore Getty Images’ massive legal surrender and lucrative partnership with OpenAI, a Chinese medical model that just beat OpenAI on its own benchmark test, Sam Altman’s new Hollywood co-production deal with director Luca Guadagnino, and the financial milestone behind Agility Robotics’ $2.5 billion public market debut.

Feedback? Email mike@yesterdayinai.news or connect on LinkedIn, X, or Bluesky. If you like the show, please take a minute to rate and review it so others can find it!

Transcript

SPEAKER_00 0:05

Hi folks, this is Yesterday in AI, your daily digest of everything happening in the world of AI in 10 minutes or less. I'm Mike Robinson. It's Thursday, June 25th, and the future just kind of fell out of the news cycle all at once. OpenAI built its own custom chip, a tech giant confessed to AI layoffs in a federal filing, Getty Images surrendered to the company it was suing, and a Chinese medical model just beat OpenAI on its own test. Let's get into it. OpenAI just announced its first custom AI silicon, a chip named Jalapeno. Designed with Broadcom, manufactured by TSMC and housed in server systems built by Celestica, prototypes are already running in OpenAI's labs right now. The plan is to have this hardware in commercial production before the end of this year. Almost every AI company is dependent on NVIDIA GPUs to serve up answers. Broadcom's CEO Hawk Tan says Jalapeno is comparable to Nvidia's Blackwell chips and Google's TPUs. OpenAI's hardware chief Richard Ho says it should hold up well for future model generations. Hyperscalers like Google and Amazon have spent years building custom processors to own their compute stacks and escape NVIDIA's pricing power. OpenAI doing this proves they are building a real infrastructure business, not just a research lab with a chat interface. With a $300 billion infrastructure deal with Oracle signed and an IPO coming later this year, controlling their silicon costs is a massive financial flag in the ground. And while OpenAI is custom building hardware to protect its balance sheet, Oracle is hacking away at its headcount to fund the infrastructure OpenAI is renting. Oracle's fiscal year 2026 annual report landed this week, and a sentence buried in the SEC filing sent a chill through the tech sector. The company stated plainly that the deployment of AI across its operations has resulted and may continue to result in workforce reductions. The numbers are staggering. Oracle eliminated 21,000 jobs over the past year, dropping its headcount from 162,000 down to 141,000. Restructuring costs shot up to $1.8 billion. While they were cutting people, they injected cash into hardware, causing capital expenditures to surge 162% to $55.7 billion. It pushed free cash flow to negative $23.7 billion. Oracle is betting its $638 billion backlog, more than half of it tied to that $300 billion OpenAI deal, will justify the spend. Tech executives usually hide behind corporate speak like restructuring or efficiency to mask layoffs. Oracle said it in plain English in a regulatory filing. Now that one major tech giant has written it into a legal document, the pressure on every other firm to admit what is happening goes way up. While Oracle is openly replacing its workers with algorithms, Getty Images is giving up a multi-year legal war to make sure its photographers' work gets fed into those same models. Getty Images announced a multi-year partnership with OpenAI, sending its stock soaring over 145% in a single day. Their licensed library of hundreds of millions of professional photos will be available inside ChatGPT for display features. This is a complete corporate flip-flop. Getty spent years acting as AI's most aggressive legal opponent. They banned AI content in 2022, sued Stability AI in 2023 for scraping 12 million images, and then watched a UK court largely reject their copyright claims last year. This isn't cynicism, it's basic survival. Image generators are threatening Getty's core licensing model, and the litigation strategy failed to stop the tech. No financial terms were disclosed, so whether Getty photographers see a dime of this cash is a chapter yet to be written. While Western creators are settling lawsuits to survive the software boom, Chinese AI firms are deploying algorithmic milestones directly into the medical sector. Two Chinese companies hit major medical AI milestones this week. First, Microport Medbot's remote surgical robot, Tumai, received the CE mark, clearing it for sale across more than 30 European countries. UK surgeons already performed the country's first telesurgery using the hardware earlier this year. Second, Chinese AI firm Beichuan released a medical model named M4. They ran it against Healthbench, OpenAI's own medical benchmark built with rubrics from 262 doctors. M4 beat GPT 5.5 by nearly 16 points on the hardest tier, keeping its hallucination rate to 3.3%. Beating OpenAI on a test they built is a massive statement. The US-China AI race is usually covered through chip bands, but the medical domain is where the stakes are human. High-performing medical AI is about to move into clinical use globally. And as AI reshapes global healthcare, OpenAI and Google are spending the week fighting for the cultural narrative inside Hollywood. Director Luca Guadagnino is collaborating with Sam Altman on a film titled Artificial, a movie about the internal tensions inside an AI lab. Major studios like Amazon, MGM, A24, Neon, and Moobi are already competing for the distribution rights. This lands the same week Google Deepmind wrote a $75 million check to A24 to build film production tools. OpenAI is positioning itself as a creative collaborator, building voice partnerships with actors, pitching Sora video generation, and co-producing content with top directors. The Writers Guild has flagged major labor risks, and 56% of U.S. adults say they prefer not to watch AI-generated scenes. But the gap between where audiences say they are and what studios are actively doing keeps widening. While Hollywood figures out how to handle digital clones, the robotics sector is graduating from the science fair straight to the public markets. Agility Robotics, the company behind the digit humanoid robot, is going public through a SPAC deal valued at $2.5 billion. Their customer list includes Amazon GXO Logistics and Toyota Manufacturing Canada. Humanoid robots are moving past the demo phase. Agility going public is a real financial event where a company deploying hardware at actual client sites must face quarterly earnings calls. They aren't alone. Germany's Nura Robotics raised $1.4 billion earlier this month, backed by Amazon and Nvidia, marking the largest venture round in German history. The humanoid sector is graduating to the capital markets. And that's it. If you have any feedback about this show, you can email Mike at yesterdayNai.news, or you can find me on LinkedIn, X or Blue Sky. And if you like this podcast and want to see it continue, please take a minute to rate and review it so others can find it. Thanks. That's all for this edition of Yesterday and AI. Stay curious, and I'll see you tomorrow.