OpenAI, Microsoft Teach AI To Teachers

OpenAI, Amazon, Meta, Microsoft, and Google are supersizing their AI budgets, with the four biggest projecting $320 billion in infrastructure spending this year. The cash is flowing into mega-scale data centers, soaring researcher salaries, and startup deals built to lock up scarce expertise.

OpenAI and partners Oracle and SoftBank are building a $60 billion data center in Texas, while Amazon is backing a 1,200-acre Indiana campus where Anthropic says it could use all 30 centers to train one system. Venture funding reached $65 billion in the first quarter, and Meta put $14.3 billion into Scale AI as Mark Zuckerberg offers up to $100 million per researcher to staff a new superintelligence lab.

Industry leaders concede the technology may fall short, but believe the greater hazard is lagging rivals. Investor Jordan Jacobs summed up the prevailing mindset: big CEOs “can’t afford to be wrong by doing too little,” fueling an arms race that shows no sign of easing.

The American Federation of Teachers will open the National Academy for A.I. Instruction in New York City with $23 million from Microsoft, OpenAI and Anthropic. The center plans to start hands-on workshops this fall to show educators how to use chatbots for lesson plans and other tasks.

Microsoft is providing $12.5 million over five years, OpenAI is adding $8 million in cash plus $2 million in technical resources, and Anthropic is contributing $500,000 for the first year. The academy will operate from the United Federation of Teachers’ downtown Manhattan headquarters and aims to foster direct dialogue between developers and educators.

The project highlights how leading tech firms are stepping in to shape classroom A.I. adoption just as Washington urges private support after freezing nearly $7 billion in school funding. The article notes educators and academics remain wary of unproven educational gains, chatbot errors and the possibility that students become lifelong customers of the sponsoring brands.

Only 10% of the workforce is proficient in AI – and their companies had a big role in making that happen. Section is bringing together top AI-focused leaders to help make sure your team doesn't fall in the 90%.

On July 17 12-4pm ET, join AI changemakers, business leaders, and enthusiasts to learn how to build an AI-native team.

Hear from heads of AI at some of the world’s leading companies on what went right in their AI deployments and how you can replicate that success in your own team at Section’s AI Strategy Summit.

Registration required. No charge to attend.

Deal volume is surging as data vendors get swallowed by larger platforms pursuing AI credibility. Databricks paid $1 billion for Neon and Salesforce spent $8 billion on Informatica in the past two months.

PitchBook counts more than $300 billion invested in over 24,000 data startups between 2020 and 2024, leaving enterprises juggling incompatible point solutions. Fivetran’s May takeover of Census illustrates incumbents stitching together end-to-end data flows to gain pricing leverage.

SnapLogic’s Gaurav Dhillon warns these pre-ChatGPT products need significant retooling before they can underpin ambitious AI rollouts. PitchBook analyst Derek Hernandez argues the real value now sits in fusing leading AI developers with data-management platforms rather than keeping them separate.

Moonvalley has opened Marey, its 3D-aware AI video generator, to all users after running it in beta since March. Access is sold through a monthly credit plan starting at $14.99 for 100 credits and topping out at $149.99 for 1,000.

The system creates clips up to five seconds long and lets filmmakers manipulate objects, characters, camera moves, and backgrounds. Moonvalley says Marey is trained entirely on openly licensed data, aiming to shield users from copyright disputes.

Director Ángel Manuel Soto reports the tool has already lowered his production costs by 20% to 40%, demonstrating immediate budget relief. The public launch puts Moonvalley head-to-head with Runway, Luma Dream Machine, Pika, and Haiper in the crowded AI video field.

A university hospital physician posted a detailed Reddit essay assessing ChatGPT’s role in medicine. He writes that many doctors already use the chatbot to double-check treatments and refresh knowledge during patient care.

He states the system excels when clinicians supply complete, objective data and already know the correct diagnosis, citing tasks like confirming vitamin B12 therapy or flagging well-documented thrombophilia cases. He adds that it misfires when key context is missing, pointing to scenarios where pregnancy or shaken-baby syndrome is overlooked because the necessary cues are absent.

The doctor concludes ChatGPT is valuable for second opinions and helping patients prepare for visits but remains unsafe as a stand-alone diagnostician or primary education source. He argues clinical judgment and in-person evaluation still determine outcomes, while the AI merely augments that process.