August 28: Nvidia's shock forecast and the supply chain gets real

Nvidia's shock 70% growth forecast resets the AI spending baseline

Nvidia beat second-quarter estimates and then, for the first time ever, gave a year-ahead forecast: 70% revenue growth in fiscal 2028, versus the roughly 44-45% Wall Street had modeled.

Why it matters

The forecast was significantly higher than Wall Street was expecting, and based on consensus fiscal 2027 revenue, sales next year would hit roughly $673 billion — putting Nvidia ahead of Apple and Alphabet, and behind only Amazon among U.S. tech companies. Analysts had never seen the company guide a year in advance before, and the magnitude of the upside speaks to how confident Nvidia is in its own forecasts. That confidence, more than the quarter itself, is what moved markets: Kress delivered the news after the market closed and her comments sent Nvidia's stock rallying more than 4% in after-hours trading.

What this means for you

Huang framed this as AI reaching "its inflection point," with tokens now productive and profitable, and demand accelerating. If Nvidia is right, the AI infrastructure buildout — and the compute budgets that ride on it — has years left to run, not months.

Finance: Nvidia flagged that its entire supply chain is challenged, with everybody running flat out — and said growth would be even higher if not for these constraints. That's a bottleneck story as much as a demand story: budget for compute costs to stay elevated and allocation to matter as much as price.

Managers: Nvidia also expanded its AWS partnership, with Amazon deploying 2 million additional Nvidia GPUs across fiscal 2027 and 2028, plus new Vera CPUs. Expect cloud AI compute pricing and availability to remain a live planning constraint into 2028, not something that eases this year.

Do this: If your team's roadmap depends on GPU-backed compute (training, inference, or cloud AI services), revisit 2027-2028 budget assumptions now — capacity and cost, not just capability, will be the constraint.

Signal 4/5· ImportantSources: Nvidia wows Wall Street with a strong quarter and an eye-popping sales forecast, Nvidia 70% growth forecast puts it on track to be tech No. 2 company

Also worth knowing

Models & capabilities

Z.ai's shares jump 8% after AI model runs entirely on Chinese chips

Z.ai released GLM-5.3-Flash, a low-cost model it says runs entirely on 100,000 domestically-made Chinese chips, and the stock surged on the news. The model briefly topped usage charts under the code name "Ox Alpha" before Z.ai confirmed it built it.

Why it matters & what to do

Why it matters

For years the assumption was that frontier-class AI needed Nvidia chips. If Z.ai's claim holds, it undercuts both the US export-control strategy and the pricing power of Western model vendors — GLM-5.3-Flash reportedly costs a fraction of comparable US models per task.

What this means for you

A credible, cheap, Chinese-chip-powered model narrows the gap between "sanctioned" and "state of the art" — expect more price pressure across the AI market.

Engineers: If you're picking models for high-volume, low-stakes tasks, GLM-5.3-Flash and its rivals are now a legitimate line item to test against your evals, not just a curiosity.

Finance: Enterprise AI subscriptions bought as "safe defaults" (OpenAI, Anthropic, Grok) may be harder to justify at scale if pay-as-you-go alternatives are this much cheaper — worth revisiting vendor contracts at renewal.

Do this: Nothing to do yet — but flag GLM-5.3-Flash for your next model-cost review if you run high-volume inference workloads.

Source: Z.ai shares surge 8% after releasing new AI model using Chinese chips

Signal 3/5· Pay attention

Tools & tactics

Anthropic launches a standard to let AI agents run lab and factory hardware

Anthropic previewed the Model Hardware Standard (MHS), a specification letting AI agents safely operate physical devices like lab equipment and robots. It's model-agnostic and cuts integration time from weeks to hours.

Why it matters & what to do

Why it matters

Connecting AI to real machines has always required custom, slow builds by specialists. Anthropic says MHS shrinks that to hours, and plans to open-source it like it did with the Model Context Protocol, which could make it a default plumbing layer for physical AI the way MCP became for data.

What this means for you

AI agents are starting to move from screens into direct control of physical equipment, a shift worth tracking even outside hardware-heavy industries.

Engineers: MHS is a framework for connecting large language models like Claude with physical objects, from manufacturing equipment to microscopes, so expect standardized agent-hardware interfaces to become part of the toolchain, not just software APIs.

Managers: Companies can integrate AI into their equipment in "hours or minutes" instead of the "weeks, if not months" it typically takes with specialist custom builds — a real cost and timeline lever if your team runs lab or manufacturing hardware.

Do this: If your org runs lab, robotics, or manufacturing equipment, ask your hardware vendors whether MHS support is on their roadmap — several (AWS, Automata, Danaher, Qiagen, Tecan among others) are already committing.

Sources: Anthropic — Previewing the Model Hardware Standard, Anthropic makes first move into physical AI with universal standard

Signal 3/5· Pay attention

Policy & risk

SK Hynix breaks ground on first US memory plant, but it's packaging, not manufacturing

SK Hynix held a groundbreaking ceremony for a $4 billion HBM packaging facility in West Lafayette, Indiana, with CEO Kwak Noh-Jung saying the site will be a "key HBM production base in America" by 2030. Chips will still be manufactured in South Korea and China and shipped to Indiana for stacking and connection.

Why it matters & what to do

Why it matters

This is the clearest test yet of whether AI's memory supply chain can actually be onshored — and the answer, for now, is only partially. Washington wants front-end fabs on US soil; SK Hynix is investing in the more capital-light, lower-risk step of packaging instead, with the cleanroom not opening until October 2028.

What this means for you

A geographically distributed AI supply chain is being built piece by piece, but the highest-value, highest-risk manufacturing step is still staying in Asia — genuine diversification will take years, not a single groundbreaking.

Finance: SK Hynix is the HBM market leader mid-way through a $720 billion buildout, with the vast majority of that expansion taking place in its home country, where the company is constructing the largest memory fab campus in the world; the Indiana plant is a hedge, not a pivot, and the stock's momentum (it just listed on Nasdaq) still hinges on Korea-based capacity.

Managers: If your roadmap assumes US-based memory supply reduces geopolitical risk, check the fine print — packaging plants don't eliminate exposure to South Korea and China-based front-end fabs.

Do this: Nothing to do yet for most readers — just be aware that "reshoring" AI hardware right now mostly means the last step of the process, not the whole chain.

Source: SK Hynix CEO says Indiana will be key memory production base by 2030, first U.S. facility now underway

Signal 3/5· Pay attention

Money & markets

Nvidia reportedly agrees to buy Hugging Face for $12.9B

Nvidia has struck a deal to acquire Hugging Face, the leading open-source AI model hub, for $12.9 billion, according to TechCrunch.

Why it matters & what to do

Why it matters

Hugging Face is where most of the industry hosts, shares and fine-tunes open models — owning it gives Nvidia a direct read on what the whole ecosystem is building next, not just the chips it runs on. The deal also marks Nvidia's re-entry into cloud services, a business it had largely ceded to AWS, Azure and Google Cloud.

What this means for you

The company that makes the GPUs now also owns the platform where open-source AI development happens — expect its tools and infrastructure to get quietly favored by default.

Engineers: If your workflow depends on Hugging Face for models, datasets or Spaces, expect tighter integration with Nvidia hardware and possibly less neutrality over time.

Finance: A vertical move like this typically triggers antitrust scrutiny and volatility in AI infrastructure stocks — watch for regulatory pushback before assuming it closes cleanly.

Do this: Nothing to do yet — just be aware, and note any Hugging Face terms-of-service changes if the deal closes.

Source: Nvidia closes in on Hugging Face acquisition

Signal 4/5· Important

One line to sound smart

“Nvidia just guided to 70% growth in 2028 and acquired Hugging Face, while Chinese chips and onshored packaging are finally showing up in production.”

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.