Daily News Coverage · October 28, 2026
The Big Headline: Nvidia-AMD $4B Meta Investment, GPT-6.1 Terra, Oracle-OpenAI Enterprise Push, GMO Extortion and a Health-System Cyberattack
Good morning, folks! Nvidia and AMD will invest $4 billion combined in Meta Platforms — $3 billion and $1 billion respectively — deepening collaboration on Meta's Muse AI agent and reviving circular-financing concerns. OpenAI ships GPT-6.1 Terra, a model that can operate a computer on its own, with reports citing roughly 3x flagship speed and an API input price cut to $1 per million tokens. Oracle expands its OpenAI partnership, integrating ChatGPT and Codex into its enterprise applications and offering Google TPUs and OpenAI's Jalapeño chips on OCI. Japan's GMO Internet refuses a $1 million ransom, and attackers begin publishing customer data. The University of New Mexico Health Sciences Center cancels classes for a second day after a cyberattack, with a group called BYOD claiming responsibility. WhiteArmor claims theft of 15.42 million Gamdom accounts and demands $15 million in Bitcoin. General Compute raises $100 million at an $800 million valuation. Physical Intelligence reportedly seeks a $20 billion valuation. Cirtek sells its semiconductor unit for $20.2 million. And Nvidia backs India's Simplismart in a $14 million Series B. Here's today's briefing from The CODEW's The Newsroom.
Nvidia and AMD to Invest $4 Billion in Meta Platforms as Muse Agent Collaboration Deepens
Nvidia will invest $3 billion and AMD $1 billion, with Meta expanding adoption of Vera Rubin, Vera CPU, Helios racks and EPYC processors — reviving concerns about circular financing.
What happened: Nvidia and AMD announced on October 27 that they will invest $3 billion and $1 billion respectively in Meta Platforms, according to Financial News. The investments accompany Meta's previously announced plan to raise up to $95 billion for AI infrastructure, including a $35 billion initial share offering. Nvidia said Meta plans to expand adoption of its Vera Rubin platform and Vera CPU, deploy Spectrum-X Ethernet and BlueField-4, and deepen collaboration on Meta's AI agent Muse. AMD said its Helios AI racks and EPYC processors will also be used for Muse. Nvidia CEO Jensen Huang described AI agents as "a new computing paradigm," while Meta CEO Mark Zuckerberg said Muse is "how we bring personal superintelligence to everyone." According to Meta's offering documents, Muse attracted more than 10 million users in the US and Canada within its first few weeks, with shopping as a primary use case. Nvidia has participated in 121 funding rounds this year, up from 13 in 2022, while AMD's participation has risen to 25 from two.
Key numbers/companies: Nvidia · AMD · Meta · $3 billion · $1 billion · $4 billion combined · Vera Rubin · Vera CPU · Helios · EPYC · Muse · 10 million users · 121 Nvidia rounds · 25 AMD rounds.
Why it matters: The investments revive concerns about circular financing — chipmakers funding their own customers' purchases. Nvidia and AMD are effectively underwriting Meta's AI buildout while selling it the silicon that powers it. For Meta, the capital injection reduces the net cost of its $95 billion raise and aligns its suppliers with its infrastructure roadmap. For the market, it signals that the largest AI buyers can now extract equity from their suppliers as a condition of procurement.
What's next: Whether the investments close alongside Meta's offering and how regulators view supplier-to-customer equity stakes.
Sources: Financial News; SETN; Nate News; CNBC.
OpenAI Ships GPT-6.1 Terra with Computer-Use Capability and Sharp API Price Cut
Reports cite roughly 3x flagship speed and an input price cut to $1 per million tokens, extending OpenAI's enterprise push one day after Anthropic filed its S-1.
What happened: OpenAI has released GPT-6.1 Terra, a model that can operate a computer on its own, according to reports compiled by NineZenith and citing Sina and MyDrivers. Multiple reports cite roughly 3x flagship speed at a sharply lower cost, with the API input price cut to $1 per million tokens. The release extends OpenAI's mid-tier pricing strategy and comes one day after rival Anthropic publicly filed its S-1 for a Nasdaq IPO. The model's computer-use capability positions it for agentic workflows that require navigating software interfaces rather than just generating text.
Key numbers/companies: OpenAI · GPT-6.1 Terra · computer use · 3x speed · $1 per million tokens · API price cut · Anthropic S-1.
Why it matters: The price cut is the most aggressive move yet in the mid-tier model war, and it puts pressure on Anthropic and Google to match on cost. The computer-use capability is also significant: it moves OpenAI's models from answering questions to operating software, which is the foundation for the agentic enterprise workflows that Oracle, Microsoft, and others are building. The timing — one day after Anthropic's S-1 — suggests OpenAI wants to define the narrative before public investors compare the two labs.
What's next: Whether the model's computer-use capability is reliable enough for enterprise deployment and how Anthropic responds on pricing.
Sources: NineZenith; Sina; MyDrivers; Tencent News.
Oracle Expands OpenAI Partnership, Integrates ChatGPT and Codex into Enterprise Apps
Oracle will also offer Google TPUs, OpenAI's Jalapeño chips and Cerebras hardware on OCI — the first time Google's custom chips run on another hyperscaler's cloud.
What happened: Oracle announced on October 27 at Oracle AI World in Las Vegas that it is expanding its partnership with OpenAI, integrating ChatGPT and Codex into Oracle's enterprise applications to make databases more actionable, according to Sina Finance. Oracle Cloud Infrastructure President Mahesh Thiagarajan said using ChatGPT "lets anyone build apps, deploy apps and manage apps." OpenAI CEO Sam Altman was scheduled to appear on stage with Oracle co-CEO Clay Magouyrk. The partnership is a further example of OpenAI's push into the enterprise market, one day after Anthropic filed its S-1. Oracle also said it is offering different hardware on OCI, including Google Tensor Processing Units, OpenAI's Jalapeño chip and Cerebras — the first time Google's custom chips will run on another hyperscaler's cloud. Thiagarajan said chip shortages remain a persistent issue and that customers benefit from having different accelerator options, while adding that Nvidia chips are "the best and most general-purpose."
Key numbers/companies: Oracle · OpenAI · ChatGPT · Codex · Sam Altman · Clay Magouyrk · Mahesh Thiagarajan · Google TPU · Jalapeño · Cerebras · OCI.
Why it matters: The integration gives OpenAI a distribution channel into Oracle's enterprise customer base, which spans healthcare, finance, and government. The hardware diversification is equally significant: by offering Google TPUs and OpenAI's custom silicon alongside Nvidia GPUs, Oracle is positioning OCI as a neutral platform for AI compute. If enterprises can choose accelerators based on workload rather than vendor lock-in, the competitive dynamics of the cloud AI market shift.
What's next: Whether enterprises adopt the integrated ChatGPT/Codex workflows and how Nvidia responds to Oracle offering competing accelerators.
Sources: Sina Finance; Ic.work; Oracle AI World.
GMO Internet Refuses $1 Million Ransom as Attackers Begin Publishing Customer Data
Five forex margin trading firms have reported damage to Japan's government, and the attackers say they will release more customer data each day at noon.
What happened: GMO Internet Group's forex trading system has been breached and is being extorted, according to NHK and TBS reports cited by Wen Wei Po. The attackers published what appeared to be detailed personal data of one customer on October 26 after GMO did not pay a $1 million ransom, and threatened to release more customer data each day at noon. GMO said the attackers initially demanded $50 million and that it will not comply, instead seeking government assistance. GMO has urged customers to run antivirus checks, isolate personal data, back up files and power down devices. Japan's Finance Minister Satsuki Katayama said five forex margin trading firms have reported damage to the government in connection with the GMO incident. Japan has seen a series of cyberattacks, including data leaks at Rakuten Group, JR Central, JR West, JR Kyushu and Okinawa rental car company OTS. Prime Minister Sanae Takaichi convened a cybersecurity cabinet meeting on October 26 and ordered emergency measures to be completed next week, including a hacker-attack detection network.
Key numbers/companies: GMO Internet · $1 million ransom · $50 million initial demand · 5 forex firms · Rakuten · JR Central · JR West · JR Kyushu · OTS · Satsuki Katayama · Sanae Takaichi · daily noon data release.
Why it matters: The GMO incident has escalated from a data breach into a sustained extortion campaign with a credible daily release schedule. The involvement of five forex firms shows that the breach is cascading into the financial sector, and the Japanese government's emergency response signals that it views the incident as a national cybersecurity event. The attackers' willingness to publish data after a refused payment sets a precedent that could encourage other groups to follow through on their threats.
What's next: Whether the attackers follow through on daily data releases and how Japan's new detection network is implemented.
Sources: Wen Wei Po; NHK; TBS.
University of New Mexico Health Sciences Center Hit by Cyberattack, Classes Cancelled for a Second Day
A group calling itself BYOD claims it compromised 36,000 machines, both Active Directory forests and two Oracle databases — claims that remain unverified.
What happened: The University of New Mexico has cancelled classes across its medicine, nursing and pharmacy faculties for a second day as it recovers from a "cybersecurity event" that hit the Health Sciences Center on Monday, October 26, according to The Register via CISO AI. The university shut down systems, cancelled appointments and clinics at UNM Health hospitals, and diverted ambulances to facilities in Albuquerque and Santa Fe. Hospital appointments and clinics were due to return to normal on the second day, though nonessential staff were asked to stay home and recovery will continue in phases. State and federal law enforcement were notified, external partners were brought in, and an Emergency Operations Center was activated. A newly formed group calling itself BYOD claimed responsibility on Monday afternoon, saying it compromised 36,000 machines, both Active Directory forests, two Oracle databases, and personal data on every student and employee. Those claims are unverified, and a group taking credit does not prove it carried out the attack.
Key numbers/companies: University of New Mexico · Health Sciences Center · BYOD · 36,000 machines · Active Directory · Oracle databases · Albuquerque · Santa Fe · Emergency Operations Center.
Why it matters: The attack on a health sciences center is a reminder that hospitals and medical schools are high-value targets with low tolerance for downtime. The diversion of ambulances shows how quickly a cyber incident can become a patient-safety issue. The BYOD group's claims, if verified, would make this one of the largest higher-education breaches of the year, though the group's newness and unverified claims warrant caution.
What's next: Whether the university confirms the scope of the breach and whether BYOD's claims are substantiated.
Sources: The Register; CISO AI.
WhiteArmor Claims Theft of 15.42 Million Gamdom Accounts, Demands $15 Million in Bitcoin
The crypto casino suffered a hot-wallet attack on October 13 with losses exceeding $3 million; no evidence yet links the two incidents.
What happened: Ransomware group WhiteArmor claims to have obtained the full user database of crypto casino Gamdom, covering approximately 15,420,003 accounts and 146 columns of data, and is demanding $15 million in Bitcoin, according to Dark Web Informer via Futunn. The group says it gained access through an internal machine account with administrator privileges and published about 100 account records as evidence. Gamdom suffered a hot-wallet attack on October 13, with confirmed losses exceeding $3 million and on-chain researchers estimating losses at about $4.2 million. There is currently no evidence linking the two incidents, and WhiteArmor's claimed breach scale has not been independently verified.
Key numbers/companies: WhiteArmor · Gamdom · 15,420,003 accounts · 146 columns · $15 million Bitcoin · $3 million hot-wallet loss · $4.2 million estimated · Dark Web Informer.
Why it matters: The Gamdom case illustrates the compounding risk facing crypto platforms: a hot-wallet exploit followed by an extortion campaign targeting user data. The claim of access through an internal administrator account, if verified, points to privileged-access management as the weak link. The $15 million demand is large relative to the confirmed hot-wallet loss, suggesting the attackers see user data as the more valuable asset.
What's next: Whether Gamdom confirms the breach and how regulators treat data extortion in the crypto sector.
Sources: Dark Web Informer; Futunn; Mitrade.
AI Compute Startup General Compute Raises $100 Million Series A at $800 Million Valuation
The round was led by Blackbird Ventures and Abstract Ventures, with the company using non-Nvidia chips and a $400 million purchase agreement with SambaNova.
What happened: AI compute startup General Compute has raised $100 million in a Series A round led by Blackbird Ventures and Abstract Ventures, valuing the company at $800 million post-money, up sharply from a $60 million valuation after its May seed round, according to TechCrunch via ChainCatcher. General Compute focuses on AI inference compute, uses specialized chips from Nvidia competitors, and has signed a $400 million chip purchase agreement with SambaNova. Other suppliers include Positron and Cerebras.
Key numbers/companies: General Compute · $100 million Series A · $800 million post-money · Blackbird Ventures · Abstract Ventures · $60 million seed valuation · SambaNova · $400 million purchase agreement · Positron · Cerebras.
Why it matters: General Compute is a direct bet that Nvidia's dominance in AI compute is not inevitable. By building inference capacity on SambaNova, Positron and Cerebras chips, it offers enterprises an alternative to Nvidia-based clouds. The $800 million valuation, up more than 13x from its seed round five months earlier, shows that investors are willing to pay a premium for credible Nvidia alternatives.
What's next: Whether General Compute can deliver inference capacity at competitive prices and how Nvidia responds.
Sources: TechCrunch; ChainCatcher.
Cirtek Sells Semiconductor Unit to India's Sancode for $20.2 Million to Pay Down Debt
The Philippine company is exiting semiconductor manufacturing after taking a hit from US tariffs, with Sancode investing a further $4.8 million in equipment and retention.
What happened: Cirtek Holdings Philippines Corp. has sold its semiconductor arm, Cirtek Electronics Corp. (CEC), to Mumbai-based Sancode Semi Private Ltd. for at least $20.2 million, according to The Philippine Star. The sale covers 155.51 million common shares representing 100% of CEC's capital stock and will be paid in two tranches: $11.32 million at closing and $8.88 million a year later. Cirtek said it will use the proceeds to pay off expiring loans, reduce debt and fund working capital at its remaining subsidiary, Cirtek Advanced Technologies and Solutions Inc. The company took a sharp hit from higher US tariffs and is reorganizing for future growth. Sancode is spending an additional $2 million to upgrade CEC's equipment and $2.8 million on an employee benefit plan to retain workers.
Key numbers/companies: Cirtek Holdings Philippines · Cirtek Electronics · Sancode Semi · $20.2 million · $11.32 million first tranche · $8.88 million second tranche · $2 million equipment upgrade · $2.8 million retention plan · US tariffs.
Why it matters: The sale is a small but telling example of how US tariffs are reshaping semiconductor supply chains in Southeast Asia. Cirtek is exiting semiconductor manufacturing to shore up its balance sheet, while an Indian buyer is acquiring capacity at a discount. The deal also shows that Indian companies are becoming more active buyers of semiconductor assets outside India.
What's next: How Sancode integrates the unit and whether other Philippine semiconductor firms pursue similar exits.
Sources: The Philippine Star.
Also in the News
Physical Intelligence seeks $20 billion valuation: Robot foundation model company Physical Intelligence is reportedly in talks for a new funding round at a valuation target of about $20 billion, according to 9fzt citing Cailianshe.
Nvidia backs Simplismart: Nvidia participated in Indian AI inference software startup Simplismart's $14 million Series B, led by Dallas Venture Capital, valuing the company at about $100 million. Simplismart says its software has helped an AI video platform cut GPU serving costs by 56%, and its annualized revenue run rate grew 14x in the past six months.
Peak XV, Lightspeed, General Catalyst in talks to back Indian AI startups: The venture firms are in advanced stages of investing in at least half a dozen Indian AI startups, with individual deals ranging from $3 million to $10 million across AI agents, biotech and enterprise software, according to Moneycontrol. Peak XV is in talks to invest $8–10 million in former Zomato COO Rinshul Chandra's new AI startup, $4–6 million in Spark Labs, and about $3 million in Erid Bio alongside General Catalyst.
Nvidia trims Vera Rubin specs: Nvidia has reduced HBM4 memory speeds and NVLink bandwidth on its upcoming Vera Rubin platform, according to Expreview via NineZenith, a sign that even next-generation racks must balance cost against memory supply.
Why It Matters
Circular financing is becoming structural in AI. Nvidia and AMD investing $4 billion in Meta while selling it the chips that power its AI buildout is the clearest example yet of suppliers underwriting their customers. This model supports demand in the short term, but it also concentrates risk: if AI spending slows, the chipmakers' equity stakes lose value at the same time their order books shrink.
OpenAI's GPT-6.1 Terra price cut is the start of a mid-tier model war. At $1 per million input tokens with 3x flagship speed and computer-use capability, OpenAI is making it expensive for enterprises to choose Anthropic or Google on cost alone. The timing — one day after Anthropic's S-1 — suggests OpenAI wants to define the public-market narrative before investors compare the two labs.
Japan's cybersecurity crisis is escalating into a government-level emergency. GMO's refusal to pay a $1 million ransom, the daily data-release threat, and five forex firms reporting damage have prompted the Prime Minister to convene a cabinet meeting and order emergency measures. The GMO case is the most serious of a wave that includes Rakuten, JR Central, JR West, JR Kyushu and OTS.
Healthcare and education remain the most vulnerable sectors. The University of New Mexico Health Sciences Center attack forced ambulance diversions and cancelled classes for two days. The BYOD group's unverified claims of compromising 36,000 machines and two Oracle databases underscore how quickly a single incident can become a regional crisis.
Investors are funding Nvidia alternatives and AI infrastructure at scale. General Compute's $800 million valuation for non-Nvidia inference capacity, Physical Intelligence's reported $20 billion target, and Nvidia's own participation in Simplismart's $14 million round all point to a market that is still willing to fund infrastructure and applied AI, even as public-market sentiment remains fragile.
The CODEW Take
Nvidia and AMD's $4 billion investment in Meta is circular financing, and the market knows it. The chipmakers are buying equity in a customer that will use the proceeds to buy their products. This is not illegal, and it is not new — vendor financing has been used in telecom, aerospace and shipping for decades. But it does mean that Nvidia's and AMD's returns are now tied to Meta's ability to monetize AI, not just to sell chips. If Meta's $95 billion raise prices poorly or its Muse agent fails to scale, the chipmakers' equity stakes lose value at the same time their order books soften.
OpenAI's GPT-6.1 Terra is a pricing weapon aimed at Anthropic's IPO. Shipping a model with computer-use capability at $1 per million input tokens, one day after Anthropic filed its S-1, is not a coincidence. OpenAI is telling public investors that it can match Anthropic on capability while undercutting it on cost. The question is whether enterprises will trust a cheaper model for computer-use workflows, where errors are expensive. If they do, Anthropic's enterprise pricing power erodes before it ever lists.
Oracle's expanded OpenAI partnership is the most important enterprise AI distribution deal of the month. Integrating ChatGPT and Codex into Oracle's ERP, HCM and SCM applications gives OpenAI a channel into thousands of large enterprises that already run Oracle. Offering Google TPUs and OpenAI's Jalapeño chips on OCI is equally significant: it makes Oracle the first hyperscaler to run Google's custom silicon on its cloud, and it positions OCI as a neutral platform for AI compute. Nvidia remains the default, but the default is no longer the only option.
GMO's refusal to pay is the right call, but the attackers are following through. Publishing one customer's data and threatening daily releases is designed to maximize pressure on GMO and on the Japanese government. The fact that five forex firms have reported damage shows that the breach is cascading beyond GMO's own systems. Japan's response — a cabinet meeting, emergency measures, a detection network — is necessary, but it will not undo the data already released. The lesson for other companies is that refusing to pay is only half the strategy; the other half is having backups, isolation and customer communication ready before the deadline.
The University of New Mexico attack is a warning about healthcare's cyber resilience. Diverting ambulances because a health sciences center's systems are down is the kind of scenario that regulators have warned about for years. The BYOD group's claims may be exaggerated, but the operational impact is real. Hospitals and medical schools need offline contingency plans that assume IT systems will be unavailable for days, not hours.
What to Watch
- Meta offering pricing: Whether the $35 billion initial raise prices at a discount and how Nvidia and AMD's $4 billion commitments are structured.
- GPT-6.1 Terra enterprise adoption: Whether enterprises trust the cheaper model for computer-use workflows and how Anthropic responds on pricing.
- Oracle-OpenAI integration rollout: When ChatGPT and Codex become available inside Oracle's ERP, HCM and SCM applications.
- GMO daily data releases: Whether the attackers follow through on their threat to publish customer data each day at noon.
- University of New Mexico breach scope: Whether the university confirms the BYOD group's claims and how recovery proceeds.
- Gamdom breach confirmation: Whether the crypto casino confirms WhiteArmor's claimed 15.42 million account exposure.
- General Compute deployment: Whether the company can deliver non-Nvidia inference capacity at competitive prices.
- Physical Intelligence round close: Whether the reported $20 billion valuation materializes and who leads the round.
- Cirtek-Sancode close: Whether the two-tranche sale closes on schedule and how Sancode integrates the unit.
- Japan's emergency cybersecurity measures: What the government announces next week and how the detection network is implemented.
Editorial Note
The CODEW Daily News Coverage tracks the most important technology developments of the day, with a focus on AI, enterprise software, cloud computing, semiconductors, cybersecurity, startups, and digital infrastructure. This edition is a pure news aggregation and coverage page — what happened — while the Daily Tech Briefing provides the deeper strategic interpretation.
Coverage is based on company announcements, public disclosures, industry reporting, and other publicly available information. Reported figures and sourced-but-unconfirmed details are noted as such.
Reviewed by Erwin Castro
on
Wednesday, October 28, 2026
Rating:
No comments: