July 15: the AI capex squeeze hits enterprise vendors

IBM's 25% one-day crash exposes the real AI capex squeeze

IBM shares fell 25%, their worst day on record, after CEO Arvind Krishna warned second-quarter results missed expectations because clients diverted spending toward AI infrastructure.

Why it matters

This isn't an AI-disruption story about software being replaced — it's a budget-reallocation story. Krishna said clients shifted their quarterly capex spend toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases. Krishna admitted IBM "did not anticipate the magnitude of the capex reprioritization," and noted clients were also distracted by cybersecurity needs tied to new AI threats. When even a 115-year-old vendor with deep enterprise relationships gets blindsided by how fast IT budgets can move, it signals every software and services vendor is exposed to the same reshuffling.

What this means for you

Enterprise vendors are now competing directly with chip and memory suppliers for the same shrinking capex pool — and losing when supply is tight. Krishna said the rising costs of chips left less room for IBM's bread-and-butter mainframes and software. If your employer or clients buy enterprise software, expect budget conversations to get sharper this year.

Finance: Cramer said the shortfall is one of the clearest signs yet that companies are reshuffling IT budgets as AI spending accelerates, with businesses prioritizing cybersecurity, hardware, and AI "tokens." Treat vendor earnings misses this quarter as a read on capex rotation, not just company-specific execution — the pattern will likely repeat across the sector.

Managers: If you're planning technology budgets, assume hardware and AI-consumption costs will keep crowding out "other" software and services line items — the category Cramer flagged as IBM's weak spot.

Do this: If you manage a vendor relationship or budget, ask now whether your supplier's roadmap depends on hardware you might be deprioritizing — don't wait for their next earnings surprise.

Signal 4/5· ImportantSources: IBM stock craters 25%, the worst day on record, after company issues second-quarter earnings warning, IBM stock tanks: CEO Arvind Krishna says AI chip spending took toll

Also worth knowing

Policy & risk

DeepMind's Hassabis wants a US AI watchdog running by year-end

Demis Hassabis is publicly calling for a new U.S. federal body to safety-test frontier AI models before release, modeled on Wall Street regulator FINRA, and wants it operational within months.

Why it matters & what to do

Why it matters

Hassabis is calling on the U.S. to establish a new AI watchdog with the power to screen the world's most advanced models and coordinate an industry-wide slowdown if dangers mount. Anthropic's Dario Amodei has separately pushed for binding rules too — the lab chiefs behind Gemini and Claude now agree Washington should regulate them, differing mainly on who holds the authority.

What this means for you

Two rival AI labs converging on "regulate us" is a signal that oversight is coming faster than the current hands-off U.S. posture suggests.

Managers: If your company builds on frontier models, expect a pre-release safety-testing regime (voluntary at first) to become a compliance checkpoint within the next year.

Do this: Nothing to do yet — just note that frontier-model deployment rules may tighten by year-end, and watch for who ends up running the body.

Source: Exclusive: Google DeepMind's Demis Hassabis calls for U.S.-led global AI watchdog

Signal 3/5· Pay attention

Policy & risk

Google Search's AI features pose "unacceptable risk" to kids, safety group finds

A new report from the Youth AI Safety Institute at Common Sense Media found Google Search's AI features, tested on minor accounts, failed to flag suicide risk, called an eating disorder symptom normal, and gave deepfake creation instructions.

Why it matters & what to do

Why it matters

Google Search is used by hundreds of millions of minors daily, making it the widest-reach consumer AI surface tested for child safety yet — unlike niche companion-app failures, this implicates the default internet gateway. Regulators and plaintiffs' lawyers now have a documented failure pattern to point to across any AI product touching consumers.

What this means for you

If your product embeds generative AI and reaches minors even incidentally, assume your safety testing will eventually be audited by an outside group and made public.

Finance: Expect this kind of report to become a recurring line item in AI-liability risk models — treat "child safety audit" as due diligence, not PR.

Managers: Get ahead of this by commissioning your own red-team audit on child-safety edge cases before a nonprofit or regulator does it for you.

Do this: If you work on or with consumer-facing AI, ask whether your product has been independently tested against self-harm, eating-disorder, and deepfake-generation prompts.

Source: Google Search AI Poses 'Unacceptable Risk' to Kids, Report Finds

Signal 4/5· Important

Policy & risk

China's AI companion rules take effect, forcing Doubao and Qwen to pull custom personas

A new Cyberspace Administration regulation on "human-like interactive AI" takes effect this week, and ByteDance's Doubao is shutting down its custom AI persona feature on July 15, directing users to a separate companion app, with Alibaba's Qwen and Tencent's Yuanbao issuing similar notices.

Why it matters & what to do

Why it matters

China is set to become the first country to impose comprehensive rules aimed at curbing the harms of anthropomorphic AI, and Beijing's rules mark the world's first attempt to regulate AI with human or anthropomorphic characteristics, according to an NYU law professor. The framework bans virtual companion and virtual relative services for anyone under 18, and requires providers to deploy age-gating, minors' modes, and guardian controls including real-time risk notifications, usage summaries, time limits, and spending caps, while self-harm conversations require manual human takeover.

What this means for you

Expect Western regulators and platforms to start referencing China's rulebook as the first working template for AI-companion safety, especially the human-takeover requirement for self-harm conversations.

Managers: If your product roadmap includes any persona-based or companion AI features, budget now for age verification, usage-time nudges, and human-escalation workflows — this is the direction global compliance is heading.

Do this: If you build or manage consumer-facing conversational AI, review whether your product would need minor-mode gating or crisis-escalation protocols under a similar rule.

Sources: ByteDance, Alibaba Pull AI Companions as Beijing Tightens Rules, China to crack down on AI chatbots around suicide, gambling

Signal 3/5· Pay attention

Money & markets

ASML raises guidance twice in one year as AI chip demand outstrips supply

ASML, the sole maker of the most advanced chipmaking machines, raised its guidance for the second time this year and reported stronger-than-expected quarterly results as its customers continue to ramp up production of AI chips, now expecting full-year sales between 43 billion and 45 billion euros, up from a prior range of 36 billion to 40 billion euros.

Why it matters & what to do

Why it matters

This isn't just a chipmaker story — it's confirmation that the AI buildout's real bottleneck has moved further upstream. Orders were "extremely strong" in the first half of the year as ASML's customers accelerated their capacity expansion plans, and the company is responding by targeting a 30% addition to its 2026 low NA EUV capacity and 30% to its Deep Ultraviolet immersion capacity. When the tool-makers themselves are racing to add capacity, memory and logic shortages look set to persist well into 2027.

What this means for you

The AI trade's most durable winners may not be the chatbot makers but the equipment suppliers nobody outside the industry names.

Finance: Guidance beats that come with capacity-expansion plans (not just backlog) signal multi-year, not cyclical, demand — worth weighting equipment and materials names alongside AI software plays.

Managers: If your product roadmap depends on GPU or memory supply, budget for continued scarcity pricing through 2027, not a near-term easing.

Do this: If you hold AI-adjacent positions, check exposure to semiconductor equipment and memory suppliers, not just chip designers — that's where pricing power currently sits.

Sources: ASML rises 4% after hiking sales forecast for second time this year on strong AI chip demand, ASML Earnings: Full-Year Outlook Raised as AI Boosts Demand at Chipmakers

Signal 3/5· Pay attention

One line to sound smart

“Enterprise IT budgets are rotating hard toward hardware and away from software as companies scramble to secure AI infrastructure before prices rise.”

Tool worth a look

Claude is the AI assistant this brief is built with — genuinely useful for drafting, summarizing dense material, and thinking through what a development actually means for you. An honest pick, not a paid link.

Try Claude →

Futureproof Daily is researched and written by AI against our editorial standards — see how we work. Sources are linked on each item. Nothing here is financial, investment, or legal advice.