Forget the Earnings Beat: Microsoft's 70% OpenAI Dependency Is the Real AI Story Today
Thursday, August 6, 2026 · 32 items · 6 min read · Updated 6:03 PM
By the Numbers
$0.20
cost per million output tokens (cheapest tier)
AI
$10B
lawsuit amount
Financial
The Day's Thesis
▶
Signal of the Day: Microsoft generated $24.1 billion in AI revenue through OpenAI in fiscal year 2026 — 70% of its total AI business — exposing a structural concentration risk that directly explains the company's pivot toward open-weight models.
The 30-Second Read:
Microsoft's $24.1B OpenAI dependency (70% of AI revenue) creates single-vendor risk unprecedented for a $3T company
1Password study: AI patching tools failed to correctly remediate software flaws 74% of the time across enterprise deployments
Glencore targets ASX secondary listing by October to access AU$4.4 trillion in superannuation assets, following 86% EBITDA surge to $10.1B
OpenAI extends unlimited text chat to free-tier ChatGPT users, compressing Microsoft's Copilot differentiation from below
Two developments today expose the same structural tension: AI's most valuable commercial relationships are also its most fragile. Microsoft's revenue concentration in a single model vendor and the persistent failure of AI security tooling at scale both point to an industry where growth has outpaced the institutional infrastructure required to sustain it.
AI & Research Frontier
Microsoft's $24.1 billion OpenAI revenue dependency — 70% of its total AI business — is a concentration risk with no precedent in the company's modern history.
That figure, drawn from a Bloomberg analysis of fiscal year data ending June 2026, reframes Microsoft's recent advocacy for open-weight models — AI systems whose parameters are publicly released — not as principle but as portfolio hedging. A company that built three decades of revenue on proprietary lock-in is now championing openness precisely because its most important product line runs on someone else's infrastructure.
The 1Password finding — AI tools failed to correctly patch software vulnerabilities 74% of the time — lands as a direct constraint on enterprise AI deployment velocity. Security teams deploying AI-assisted remediation at scale are, by this measure, introducing more noise than resolution into their patch pipelines.
Suno’s AI vocalists could learn a thing or two from this glitchy image. | Image: Cath Virginia / The Verge, Getty Images
Suno announced plans to implement a new watermarking technology and download policy to limit the spread of spammy AI tracks and increase transparency. In a lengthy blog post, CEO and co-founder Mikey Shulman laid out the company's principles and the next steps for the company as it seeks legitimacy.
The company is rolling out new transparency tools, along with new watermarking and fingerprinting tech. Shulman says it aligns with "emerging industry standards" that will make it easier to identify Suno-generated content. He also says the company is aiming to partner with "distribution platforms on combatting fraud and misuse."
Last year, th …
Read the full story at The Verge.
The enterprise confidence gap flagged in the running storyline widens further: IBM's 92% access-control vulnerability rate and 1Password's 74% patching failure rate now form a consistent empirical pattern, not isolated data points.
Google's internal reorganization, with a new Gemini model team formed as commercial AI investments accelerate, reflects the same pressure: frontier research structures are being reshaped around revenue-generating product lines.
Technology & Infrastructure
OpenAI's removal of text chat limits for free and Go-tier ChatGPT users next week directly undercuts Microsoft Copilot's value proposition at the entry level.
Unlimited free text access — combined with a new "Think" button giving free users access to higher-reasoning queries — collapses the functional gap between OpenAI's $0 tier and Microsoft's paid Copilot integrations. For Microsoft, which derives 70% of its AI revenue from OpenAI, this is a distribution decision made by its primary supplier that shapes downstream pricing power.
The SemiWiki white paper published today on AI data center semiconductor architecture provides context: the inference compute — semiconductor workloads that serve user queries rather than train models — required to support unlimited free-tier usage at ChatGPT's scale is substantial, implying OpenAI is absorbing significant marginal cost to expand addressable users.
Suno's announcement of watermarking and fingerprinting technology for AI-generated music, explicitly aligned with "emerging industry standards," marks a concrete provenance infrastructure step for generative audio — a $0 marginal-cost content category where fraud detection is now a platform survival requirement.
The FCC's 2-1 vote eliminating the 39% national broadcast ownership cap — the rule restricting any single owner from holding stations reaching more than 39% of U.S. TV households — restructures the addressable market for AI-driven content distribution and targeted advertising at scale.
Markets & Capital Flows
Glencore's planned ASX secondary listing, targeting October, is a direct capital access play on Australia's AU$4.4 trillion superannuation sector — which is projected to reach AU$12.4 trillion by 2045.
The listing requires no new share issuance and will execute via CHESS Depositary Interests (CDIs — ASX-listed instruments representing shares in a foreign company), with CEO Gary Nagle explicitly citing pension fund mandates that restrict overseas allocations as the structural barrier being removed.
Glencore's 86% EBITDA surge to $10.1 billion in H1 2026 — driven by energy market volatility from Middle East conflict escalation — provides the earnings credibility to clear the AU$1.5 billion local market-cap threshold required for S&P/ASX 200 inclusion within 12 months.
Treasury yields rose Thursday as traders priced rising rate expectations, tightening the discount rate environment for long-duration infrastructure and mining capex. Ukraine's drone strike on the Slavneft-Yanos refinery in Yaroslavl — one of Russia's largest — reinforced the refining capacity constraint that analysts cite as the structural reason gas prices will remain elevated into autumn even if crude stabilizes.
Fox's advertising revenue surged 78% in the quarter, almost entirely attributable to FIFA Men's World Cup rights and Tubi growth, demonstrating that live sports rights retain pricing power in an otherwise fragmented video market.
Critical Minerals & Supply Chain
Glencore's $4 billion H1 copper capex — targeting 1 million metric tons of annualized production by 2028 and 1.6 million by 2035 — is the largest single disclosed copper expansion program active in the market today.
The DRC regulatory friction at Kamoto (190,000 metric tons annually, 70% Glencore stake) did not interrupt production following President Tshisekedi's intervention, but the episode illustrates the jurisdictional risk embedded in the copper supply build that AI data center infrastructure — which requires substantial copper for power distribution and cooling — ultimately depends on.
Guinea's Nimba Mining Company securing its mining convention by government decree establishes the legal framework for bauxite extraction and export from Port of Kamsar, adding a new supply-chain node to the aluminum input market. Bauxite-to-alumina-to-aluminum is a three-step refining chain that feeds both EV battery casings and data center structural materials.
The Interconnect: Cross-Sector Causal Chains
→Microsoft's 70% OpenAI revenue concentration (fiscal year ending June 2026) → creates financial incentive to support open-weight model ecosystems as a hedge against single-vendor pricing power → enterprise AI infrastructure spend diversifies away from pure OpenAI-stack deployments toward multi-model architectures, increasing semiconductor demand for varied inference hardware reported
→1Password's finding that AI patching tools fail 74% of the time → enterprise security teams slow autonomous AI deployment in production code pipelines → slows the velocity of AI-driven developer tooling adoption, constraining near-term software infrastructure capex growth for vendors dependent on AI code-assist monetization reported
→Glencore's $4B H1 copper capex targeting 1.6 million metric tons by 2035 → expands long-dated copper supply into the period when AI data center buildout reaches peak power infrastructure demand → ASX secondary listing in October unlocks AU$4.4 trillion superannuation capital to finance that expansion, reducing Glencore's dependence on European debt markets reported
Watchlist
▸Microsoft — OpenAI revenue concentration and open-weight model strategy evolution · Catalyst: Next quarterly earnings disclosure and any OpenAI commercial agreement renegotiation · When: Q1 FY2027 earnings, October 2026
▸OpenAI — Free-tier unlimited chat rollout impact on Copilot pricing pressure and Microsoft revenue share · Catalyst: ChatGPT free-tier unlimited text launch · When: Week of August 11, 2026
▸Glencore — ASX secondary listing execution and S&P/ASX 200 inclusion timeline · Catalyst: CDI listing filing and October target date · When: October 2026
▸Glencore / DRC — Kamoto Copper production continuity amid ongoing tax dispute · Catalyst: Resumption or collapse of settlement talks with Congolese revenue authority · When: Ongoing; next cabinet review flagged post-July 10 intervention
▸1Password / Enterprise AI Security — Follow-on vendor responses to 74% AI patching failure rate finding · Catalyst: Enterprise security procurement cycles and competing study releases · When: Q3 2026
▸Google DeepMind — New Gemini model team structure and capability output post-leadership reorganization · Catalyst: First model release or benchmark disclosure under new team structure · When: Q4 2026
▸FCC — Broadcast consolidation activity following removal of 39% national ownership cap · Catalyst: First major ownership transaction filed under new rules · When: 60–90 days post-ruling
Glencore (LSE:GLEN,OTCPL:GLCNF) will pursue a secondary listing on the Australian Securities Exchange (ASX) as early as October, attempting to tap into the country's US$3.1 trillion pension pool following a massive surge in first-half corporate profits.
Chief Executive Gary Nagle said the company is targeting inclusion in the benchmark S&P/ASX 200 (INDEXASX: XJO) within 12 months. That initial threshold requires about AU$1.5 billion (US$ 1.1 billion) in local market capitalization.
The listing requires no new capital raising and will execute through CHESS Depositary Interests. Nagle initiated the strategy following direct appeals from major Australian pension funds, whose mandates restrict overseas allocations.
Australia's superannuation sector currently holds AU$4.4 trillion (US$ 3.1 trillion) in assets, a figure projected to hit AU$12.4 trillion (US$8.7 trillion) by 2045.
“They have said to us that if there was an ASX line, that they’d be able to invest a lot more in Glencore,” Nagle reportedly told investors during a half-year results call, while adding that the Australian market offers a sophisticated investor base with deep expertise in global resources and commodity-cycle volatility.
The Australian expansion coincides with the expiration of a six-month standstill on merger discussions with Rio Tinto (ASX:RIO,NYSE:RIO,LSE:RIO). While Nagle stated a secondary listing would not alter the mechanics of a potential merger, he acknowledged the previous talks drove local investor interest.
The listing push follows a windfall first half for the Swiss commodity trader, driven by extreme volatility in global energy markets following the escalation of the Middle East conflict.
Glencore reported an 86 percent increase in Group Adjusted EBITDA to US$10.1 billion. Net income attributable to equity holders jumped by more than US$5 billion to reach US$4.4 billion.
Concurrently, the company deployed US$4 billion in first-half capital expenditure to secure land access and operational flexibility across its copper portfolio. Glencore aims to reach 1 million metric tons of annualized copper production by 2028, climbing to 1.6 million tons by 2035.
While Glencore courts Australian capital to fund those copper ambitions, it faces regulatory friction in central Africa.
On July 9, Congolese tax authorities sealed the Kolwezi offices of Glencore's Kamoto Copper subsidiary over a multibillion-dollar payment dispute. The raid occurred after settlement talks between the company and the state revenue agency collapsed.
The escalation forced DRC President Felix Tshisekedi to intervene. During a July 10 cabinet meeting, Tshisekedi instructed the finance and mining ministries to stop unpredictable bank account seizures and asset freezes, ordering the immediate removal of police and soldiers from mining sites.
The dispute did not disrupt physical extraction at the Kamoto complex, where Glencore holds a 70 percent stake and produces roughly 190,000 metric tons of copper annually.
Don’t forget to follow us @INN_Resource for real-time news updates!
Securities Disclosure: I, Giann Liguid, hold no direct investment interest in any company mentioned in this article.
Microsoft generated $24.1 billion in AI revenue through OpenAI in the fiscal year ending in June. That's about 70 percent of its total AI business, according to a Bloomberg analysis. The heavy reliance helps explain why a company long known for vendor lock-in has recently been championing open-weight models and pushing back against proprietary isolation.
The article Microsoft's AI revenue reportedly depends on OpenAI for 70 percent appeared first on The Decoder.
The era of set broadcast ownership limits is officially over, after the Federal Communications Commission (FCC) voted Thursday to end the national ownership cap rule.
The agency's two Republicans, Chair Brendan Carr and Commissioner Olivia Trusty, voted to end the ownership cap, which restricts broadcast owners from holding stations that reach a combined more than 39 percent of US TV households, while Democratic Commissioner Anna Gomez dissented. It formalizes a policy Carr has long criticized, and which he announced last month that he would seek to end at Thursday's open FCC meeting. In place of a set limit, the FCC says, there will now be …
Read the full story at The Verge.
The U.S. appeared to quickly reject the terms of an Iranian draft plan that would open the Strait of Hormuz to ships except those from the U.S. and Israel.
Solvonis Therapeutics plc (LSE: SVNS), a late clinical-stage biopharmaceutical company developing novel small-molecule therapeutics for high-burden central nervous system ("CNS") disorders, is pleased to announce encouraging results from the initial in vitro cardiac ion-channel and broader off-target screening of SVN-015 under the U.S. National Institute on Drug Abuse's ("NIDA") Addiction Treatment Discovery Program ("ATDP"). The Company previously announced SVN-015's acceptance into the ATDP in December 2025.
Following review of the initial screening results, NIDA has confirmed that SVN-015 will advance into further evaluation under the ATDP. This is expected to include confirmatory transporter studies and in vivo studies to characterise the onset and duration of its pharmacological activity.
SVN-015 is a proprietary discovery-stage small molecule designed to modulate key monoamine transporters, including the dopamine transporter ("DAT") and serotonin transporter ("SERT"). It is being developed initially as a potential treatment for stimulant use disorder, including cocaine and methamphetamine use disorders.
Encouraging Initial Screening Results
The initial NIDA evaluation was designed to identify potential cardiac ion-channel and broader off-target liabilities before SVN-015 progressed into more extensive preclinical pharmacological studies.
SVN-015 completed the initial cardiac ion-channel screening and demonstrated an encouraging broader off-target profile. The results were sufficiently encouraging for NIDA to progress the compound into further evaluation.
On the cardiac ion-channel measures assessed to date, SVN-015 demonstrated a more favourable profile than GBR-12909, an earlier dopamine transporter inhibitor evaluated by NIDA as a potential treatment for cocaine dependence and known to interact with several cardiac ion channels.
These preliminary preclinical findings do not establish the safety of SVN-015 in humans or whole animals. Further safety pharmacology, toxicology and clinical evaluation will be required as the programme progresses.
Advancement into Further Preclinical Evaluation
The next stage of the NIDA programme is expected to further characterise SVN-015's intended transporter pharmacology, off-target profile and activity in vivo.
The planned work is expected to include:
in vivo mouse locomotor-activity time-course study to characterise the onset and duration of SVN-015's pharmacological effects
confirmatory DAT binding and functional studies;
follow-up assessment of activity at the 5-HT2B receptor.
These studies will be funded and undertaken through NIDA's preclinical ATDP. This support provides Solvonis with access to NIDA-funded specialist research capabilities rather than a grant payment to the Company. Solvonis retains ownership of SVN-015 and its associated intellectual property.
Successful completion of this stage would provide Solvonis with an important preclinical data package to inform the compound's future development strategy and could support applications for additional non-dilutive NIH or NIDA development funding.
SVN-015 and the Unmet Need in Stimulant Use Disorder
There is currently no FDA-approved medication for stimulant use disorder, including cocaine and methamphetamine use disorders. Existing treatment is primarily based on behavioural interventions, leaving a significant unmet need for effective pharmacological treatment options.
SVN-015 emerged from Solvonis' proprietary, AI-enabled CNS discovery programme. Composition-of-matter patent applications have been filed to protect the compound and related intellectual property.
Anthony Tennyson, Chief Executive Officer of Solvonis Therapeutics, said: "This is an important development milestone for SVN-015. The compound has completed NIDA's initial cardiac ion-channel and broader off-target screening, demonstrated an encouraging initial profile and will now advance into further NIDA-funded evaluation.
"Particularly encouraging is SVN-015's cardiac ion-channel profile relative to GBR-12909. These early results suggest that SVN-015 may have the potential to deliver the intended transporter pharmacology while avoiding some of the cardiac ion-channel liabilities associated with previous compounds in this area.
"NIDA's decision to progress SVN-015 provides important external validation of the case for its further evaluation and gives Solvonis access to specialist, non-dilutive preclinical development capabilities. The next stage should tell us considerably more about the compound's transporter pharmacology, selectivity and activity in vivo."
Professor David Nutt, Chief Scientific Officer of Solvonis Therapeutics, added: "These are highly promising early findings. Cardiac ion-channel activity has been an important challenge in the development of previous dopamine transporter inhibitors. Across the cardiac ion-channel assays completed to date, SVN-015's profile appears materially more favourable than that of GBR-12909.
"The next stage of NIDA's evaluation will be important in confirming the compound's transporter pharmacology, further characterising its off-target profile and establishing the onset and duration of its pharmacological effects in vivo.
"There remains an urgent need for effective pharmacological treatments for cocaine and methamphetamine use disorders, and we are very pleased that SVN-015 is progressing into this next stage of evaluation."
The content of this press release is solely the responsibility of Solvonis Therapeutics and does not necessarily represent the official views of the U.S. National Institutes of Health or the U.S. National Institute on Drug Abuse.
Enquiries
Solvonis Therapeutics plc
Anthony Tennyson, CEO & Executive Director
info@solvonis.com
Singer Capital Markets (Broker)
Russell Cook
+44 (0) 20 7496 3000
About Solvonis Therapeutics plc
Solvonis Therapeutics plc (LSE: SVNS) is a late clinical-stage biopharmaceutical company developing small-molecule therapeutics for high-burden central nervous system ("CNS") disorders. Headquartered in London and listed on the Main Market of the London Stock Exchange, Solvonis is advancing a differentiated pipeline of repurposed and discovery-stage compounds across addiction and psychiatry.
The Company's lead programmes target Alcohol Use Disorder ("AUD") and Post-Traumatic Stress Disorder ("PTSD"), with additional development and discovery work supporting expansion into further addiction and psychiatric indications, including stimulant use disorder and depressive disorders.
Its lead asset, SVN-001, is currently in Phase 3 for severe AUD in the UK, while SVN-002 is being advanced towards a planned Phase 2b trial in the United States targeting moderate-to-severe AUD. The Company's PTSD discovery programme has identified SVN-114 as a lead compound, emerging from a proprietary compound series designed to modulate key brain-signalling systems associated with emotional processing and social behaviour.
SVN-015 is the Company's proprietary discovery-stage candidate targeting stimulant use disorder. Following encouraging initial cardiac ion-channel and broader off-target screening results, it is advancing into further preclinical evaluation under NIDA's Addiction Treatment Discovery Program.
In parallel, Solvonis is advancing proprietary CNS discovery programmes supported by a dedicated compound library and AI-enabled discovery capabilities to identify new small-molecule modulators of key neurotransmitter systems.
solvonis.com | LinkedIn |X
Source
Artificial intelligence AI has become one of the most significant drivers of semiconductor innovation in history. While AI applications often receive the most attention, the underlying hardware infrastructure that enables AI consists of a highly sophisticated ecosystem of semiconductor technologies. Modern AI data centers… Read More
The post White Paper: The Semiconductor Foundation of Modern AI Data Centers appeared first on SemiWiki.
OpenAI is making a big change for ChatGPT users on its free and Go tiers: starting next week, users on those tiers will be able to have unlimited text chats with the chatbot, according to OpenAI. Right now, you may run into rate limits if you do too many text chats on those tiers, but those limits, just for text, will be going away soon. Messages that include things like file uploads and images will continue to have limits.
Next week, OpenAI is also adding a "Think" button for free and Go users that they can use to "access higher reasoning for harder questions." This week, for those users, the company is additionally upgrading ChatGPT's def …
Read the full story at The Verge.
Guinea bauxite miner Nimba Mining Company (NMC) reports that the signing of its mining convention has been concluded, establishing a clear and stable legal framework for the company to conduct activities across the value chain, from the extraction of bauxite at the Tinguilinta mine to its export from the Port of Kamsar. The mining convention, granted by decree, sets out the reciprocal rights and obligations of the Guinean State and the company for the full duration of operations. It secures the applicable regime to NMC's activities in fiscal, customs, environmental and social matters and gives the company the required visibility to commit to its long-term investments, starting with the ramp-up of the Tinguilinta site and the studies on local bauxite processing.
Stanford offers a free online AI course taught by Peter Norvig and Sebastian Thrun that focuses on foundational concepts beyond current tools like ChatGPT. This addresses a gap in practical AI education by providing deep theoretical grounding that practitioners need to understand how modern systems actually work.
GlobalFoundries' data center growth is shifting U.S. photonics from a subsidy-dependent pitch to a critical AI infrastructure bottleneck. Photonics capacity now directly constrains data center expansion, creating a measurable supply chain inflection point for semiconductor-adjacent optical components.
Eli Lilly is widening its lead over Novo Nordisk in the obesity drug market as Novo struggles to regain market share and investor confidence. The divergence signals a market leadership shift in GLP-1 pharmaceuticals with significant implications for competitive positioning and pipeline risk.
Bitcoin faced risk-off sentiment, institutional redemptions, and corporate treasury stress in Q2, creating downward price pressure, but on-chain metrics suggest potential for long-term recovery. This volatile period reflects broader institutional adoption and macro uncertainty affecting digital asset valuations.
Meta's Llama-based Muse Spark 1.2 and Muse Code agent now compete on price with a minimum tier at 20 cents per million output tokens, requiring data-sharing for training. This represents a shift from performance competition to cost-based commoditization in open-weights AI models.
Cloudflare open-sourced an AI agent workspace platform originally built for internal employee use. This democratizes agentic AI tooling and expands developer access to workflow automation infrastructure.
Trump won a temporary reprieve blocking BBC access to his business records in a $10 billion defamation lawsuit related to a 2024 'Panorama' documentary about January 6. This is a legal procedural development with no direct technology, AI, or supply chain impact.
Deutsche Bank and KBC Group froze Radiant World accounts while other banks suspended credit lines after Bloomberg reported the iron ore trading house provided falsified documents to financial institutions. The coordinated banking action reflects systemic risk concerns and compliance failures in commodity trading infrastructure.
DailySand