Sources: Nvidia has reworked a deal to finance an OpenAI Ohio data center campus so that it would initially guarantee only half of its planned $250B backstop
THE SO WHAT
Dialing a $250B, 10‑GW backstop down to roughly half is a reminder that even hyperscale AI buildout has capital pacing and risk limits. Infra-heavy AI projects should expect more staged commitments, milestone gates, and renegotiations—structure your own long-term deals so they can flex without collapsing.
READ THE SOURCE
MORE FROM THE WIRE
Applied AISources: Anthropic projects 2028 revenue of ~$190B-$200B, vs. a revenue run rate of $47B in May, as bankers and investors price the company ahead of an IPO
A ~$47B run rate with a 4x+ revenue target by 2028 is Wall Street underwriting AI infra and assistant spend as a core utility, not a side bet. For operators, this means sustained pricing power for top models and continued pressure to either commit to a major lab or build credible in-house alternatives.
Applied AI'There's no realistic chance all of that content will get remastered from scratch': Dolby Vision 2 will come to more titles than you might expect, thanks to an automated process — but there will be two key creative-led features that aren't automated
Automated Dolby Vision 2 upgrades turn a back catalog problem into a pipeline problem—streamers can raise perceived quality across thousands of titles without full remasters. Expect more tooling where AI handles bulk technical passes while a small creative team reserves judgment for the last 5% that actually moves the needle.
Applied AIGLM-5.3 is here with advanced cyber capabilities — and reportedly already found a 'serious vulnerability' in Cursor
Frontier-grade cyber capability is leaking into largely open models — if GLM-5.3 can autonomously surface “serious” vulns in popular tools, the offense–defense gap compresses for everyone at once. Treat model selection as a security decision now: you need policies for who can run what models against your own code and infra.
Applied AIZ.ai says GLM-5.3 scores 84.5% on CyberGym, vs. Mythos 5's 83.8%, and its most sensitive cybersecurity functions will only be available to verified users
We’re getting a template for gated offensive-grade AI — GLM-5.3 is open source, but its highest-sensitivity cyber tools sit behind verification, while matching Mythos 5 on CyberGym at 84.5%. Enterprises building or deploying security-capable models should expect regulators and customers to ask for similar tiered access and audit trails.