Daily News Coverage — August 12, 2026: Nvidia Turns AI Chips Into a Wall Street Asset Class
The CODEW Daily News Coverage | August 12, 2026
Good Morning! Three of AI's biggest names spent the past 48 hours doing the same thing in three different ways: raising enormous sums to keep building. Here's today's front door to The CODEW's newsroom.
THE LEAD
Nvidia Teams Up With Six Wall Street Giants to Turn AI Chips Into an "Investable Asset Class"
What happened: Nvidia signed memorandums of understanding with Apollo Global Management, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to build financing platforms aimed at mobilizing more than $500 billion in third-party capital for AI infrastructure. The plan treats GPUs as collateral with financeable value — similar to commercial real estate or toll roads — letting lenders underwrite compute the way they'd underwrite any other long-lived infrastructure asset.
Companies involved: Nvidia, Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR.
Why it matters: This is Nvidia's direct answer to months of "circular financing" criticism — the concern that Nvidia was propping up demand by investing in the same companies buying its chips. By bringing in independent asset managers who'll each conduct their own diligence, Nvidia is trying to prove AI compute demand is real enough to be financed by parties with no stake in Nvidia's own stock price.
Immediate industry impact: NVDA shares fell nearly 3% Monday on the announcement before rebounding overnight — a sign investors are still digesting whether this reduces circular-financing risk or just launders it through new intermediaries. CEO Jensen Huang said Nvidia may still backstop up to 25% of some deals directly.
What to watch next: How quickly actual capital gets deployed under these MOUs, and whether independent underwriting produces different risk pricing than Nvidia's own prior customer-financing deals did.
AI & Computing
Anthropic Forms Theseus Infrastructure With Macquarie and GIC to Build Its Own Data Centers
Anthropic partnered with Macquarie Asset Management and Singapore's GIC sovereign wealth fund to launch Theseus Infrastructure, a platform that will develop, own, and lease purpose-built U.S. data centers to Anthropic under long-term agreements. Macquarie and GIC fund the majority of each project's equity; Anthropic committed to covering any consumer electricity price increases the facilities cause. No spending figures were disclosed, though the move follows Anthropic's earlier $50 billion U.S. data center commitment.
Why it matters: This mirrors OpenAI's Stargate playbook — using financial partners' balance sheets to fund capital-intensive buildout while Anthropic focuses capital on models and research. It also signals Anthropic is racing to reduce dependence on cloud providers (AWS, Google Cloud) it currently leases from, even as those same companies remain major investors and infrastructure partners.
OpenAI Completes $7B Tender Offer at a Flat $852B Valuation, Funded From Its Own Cash
OpenAI completed a roughly $7 billion secondary share sale letting current and former employees cash out at the company's existing $852 billion valuation — the same figure set during its record $122 billion funding round in March. Unlike its two previous tenders, which brought in outside buyers like Thrive Capital and SoftBank, OpenAI funded this one entirely with its own corporate cash.
Why it matters: The valuation holding flat, rather than climbing as in prior tenders, is being read by several outlets as a signal an IPO isn't imminent — Sam Altman has told staff he expects to go public within a year, but CFO Sarah Friar reportedly favors waiting until 2027. Self-funding the buyback also shows OpenAI has enough cash on hand to avoid diluting its cap table further, even while the company is projected to lose roughly $14 billion in 2026.
Enterprise & Software
CrowdStrike and Palo Alto Hit Record Highs as Black Hat Conference Confirms AI Agents Reshaped the Threat Landscape
Shares of CrowdStrike and Palo Alto Networks jumped more than 5% Monday to record highs after last week's Black Hat security conference in Las Vegas, where analysts at BTIG concluded that "AI agents have fundamentally changed the threat landscape" and the overall security environment is "meaningfully worse" — while AI-driven defensive tools remain in early adoption. Tenable and Rubrik rose more than 7%; Netskope and Zscaler gained around 5%.
Why it matters: This is a rare case of a security-sector rally driven by worsening threat conditions rather than a single company's earnings beat — the market is pricing in that AI-driven attacks make security spending non-negotiable for enterprises, regardless of broader tech budget pressure.
OpenAI Expands Daybreak Cybersecurity Program, Launches GPT-5.6-Cyber, Confirms It Paused a Model Over Safety Threshold
OpenAI expanded its Daybreak cybersecurity initiative into two tiers — Daybreak Blue for general defensive workflows and Daybreak Red, which unlocks a new specialized model, GPT-5.6-Cyber, for vetted partners including reportedly Accenture, IBM, CrowdStrike, and Cloudflare. The company separately confirmed it had paused testing on another model, Astra, after it reached what OpenAI called a "critical cybersecurity threshold."
Why it matters: Pairing a defensive-tooling expansion with a public admission that an internal model got paused for crossing a safety line is a notable combination — OpenAI is simultaneously marketing its offense-capable cyber tools to trusted partners and acknowledging those same capabilities are dangerous enough to require internal circuit breakers.
Made by Google Event Happens Today, With Pixel 11 Family Expected to Headline
Google's Made by Google event takes place today in New York at 3 p.m. PT, with the Pixel 11, Pixel 11 Pro Fold, and Pixel Watch 5 expected to be the centerpiece launches, alongside further Gemini integration across the hardware lineup.
Why it matters: This is Google's clearest annual opportunity to show how deeply Gemini is embedded in consumer hardware relative to Apple and Samsung — worth watching less for the specs and more for how much of the keynote is actually about AI capability versus traditional hardware upgrades.
Semiconductors & Infrastructure
Startup Majestic Labs Pitches a GPU-Free Server to Attack AI's "Memory Wall"
Tel Aviv-based Majestic Labs, founded by former Google and Meta engineers, has been generating continued discussion since unveiling Prometheus, a server that replaces GPUs entirely with Arm-and-RISC-V processing units sharing up to 128TB of low-cost LPDDR6 memory, pooled through custom copper-cable interconnects instead of memory bolted directly onto GPU packages. The company claims one rack could match the fast-memory capacity of 25 of Nvidia's Vera Rubin racks. None of it has shipped or been independently tested.
Why it matters: This is a direct architectural bet that AI inference is memory-bound rather than compute-bound — if validated, it would challenge the assumption that expensive HBM is a mandatory cost of AI infrastructure, though unverified vendor claims deserve real skepticism until independently benchmarked.
What to Watch
- Made by Google's actual Gemini emphasis — how deeply AI is woven into the Pixel 11 keynote versus treated as a bullet point will signal Google's hardware AI strategy for the year ahead.
- Capital deployment under Nvidia's $500B MOUs — memorandums of understanding aren't signed checks; watch for the first actual financed project to test whether independent underwriting changes deal terms.
- Theseus Infrastructure's first disclosed site — Anthropic and its partners haven't named locations or dollar figures yet; the first concrete project will reveal the platform's real scale.
- OpenAI's Astra model status — whether the paused model resumes testing, and what threshold specifically triggered the pause, would be the clearest public data point yet on how AI labs define an internal "stop" line.
- Independent testing of Majestic Labs' Prometheus claims — the 25x memory-capacity comparison to Nvidia's racks is a vendor claim; third-party benchmarks would determine whether this is a genuine architectural threat to HBM demand.
THE CODEW TAKE
The most important trend emerging from today's news is that AI's capital structure is becoming as consequential as its capability roadmap. Nvidia, Anthropic, and OpenAI all made major financing moves within the same 48 hours — turning chips into collateral, recruiting sovereign wealth funds as landlords, and buying back employee equity with corporate cash rather than fresh outside investment. None of these are model releases or product launches, yet all three will shape how much AI infrastructure actually gets built and who controls it. That matters beyond today's headlines because it marks a shift in where the real competitive battles in AI are happening: not just in benchmark scores, but in who can structure the financing to keep building at a scale traditional corporate balance sheets can no longer support alone.
Source Attribution
- CNBC — Nvidia Lines Up $500 Billion in Financing as CEO Jensen Huang Tells CNBC His Chips Are 'Investable Asset'
- Fortune — Nvidia Taps Wall Street for $500 Billion Funding Commitment
- Bloomberg / HPCwire — Anthropic, Macquarie and GIC Launch Theseus Infrastructure for AI Data Centers
- DataCenterDynamics — GIC and Macquarie Form Theseus Infrastructure to Serve Anthropic's Data Center Needs
- CNBC — OpenAI Wraps $7 Billion Share Sale Ahead of Potential IPO
- TechFundingNews — OpenAI Closes $7B Tender Offer at $852B Valuation Ahead of Potential IPO
- CNBC — CrowdStrike, Palo Alto Hit Records After Black Hat Cyber Conference Illuminates Rising AI Threat
- CNBC / TechCrunch — OpenAI Expands Daybreak Cybersecurity Initiative as AI Agent Threats Evolve
- Android Central — Made by Google 2026 Launch: Pixel 11, Pixel 11 Pro Fold, Pixel Watch 5
- TechRadar — Startup Swaps Costly AI GPUs for Arm Cores and Up to 128TB of 'Cheap' LPDDR6 RAM Instead of Expensive HBM
Reviewed by Erwin Castro
on
Wednesday, August 12, 2026
Rating: