Sanctity.AI

Sanctity.AI Building a world in which use of AI is safe, reliable, accessible, responsible, and inviolable

TSMC just printed its strongest Q1 on record. Revenue up 35% to $35.7B. 2nm capacity booked through 2028. CoWoS packagin...
20/04/2026

TSMC just printed its strongest Q1 on record. Revenue up 35% to $35.7B. 2nm capacity booked through 2028. CoWoS packaging on track to quadruple by year end. $56B of capex pencilled in for 2026.

The obvious read: more fabs, more Nvidia, more Apple, same story with bigger numbers.

Look underneath and the story changes. If every serious AI buyer in the world already has a reservation through 2028, the bottleneck on the next wave of AI capacity is no longer the fab. It is the equipment that builds the fab. ASML, Applied Materials, KLA, and Lam Research become the vendors who decide how fast the whole industry can grow. Intel and Samsung get a quiet re-rating too, because if equipment is the constraint, second-tier fabs can spend their way toward parity.

Go one layer deeper. Taiwan imports over 95% of its energy. Every chip in this supply chain runs on a grid that runs on tankers. A Strait of Hormuz disruption is not just a foreign-policy story. It is an AI-scaling story.

The AI capital stack has models at the top and electrons at the bottom. The further down you go, the more power each layer has over the one above it.

The binding constraint on AI is not the model anymore.
One fab. One island. One strait.

Which layer are you watching?
ai breaks down the AI stack one layer at a time. Follow for the next one.

18/04/2026

If you build AI infrastructure, the bottleneck moved.
It is not chips. It is not data. It is a grey steel box with copper windings: the high-voltage transformer.
Half of planned US AI data centers for 2026 are blocked or delayed. Only 5 GW of 16 GW planned capacity is under construction. The grid queue in ERCOT is 226 GW (quadrupled in a year). Wait times for a transformer have tripled since 2019, from 24 months to 60. China XD and TBEA have order books filled through 2027.
So hyperscalers stopped waiting.
Meta is building Hyperion (7.46 GW, gas, behind-the-meter with Entergy) and Socrates. Microsoft has a 1.4 GW West Virginia LOI plus 2.5 GW in talks with Chevron in West Texas. AWS already takes 1.9 GW behind-the-meter from Susquehanna nuclear. NGI forecasts 35 GW of behind-the-meter AI capacity by 2030.
This is not a workaround. It is a model inversion.
When the grid is a queue, power is a line item. When every hyperscaler owns its own generation, power becomes the moat. Compliance follows the meter. The regulated utility relationship turns into a private industrial deal. A company that owns its silicon, its models, its fibre, and now its electrons is not subject to the same regulatory geometry as a tenant of the public grid.
Three questions worth holding before the next build.
Price power like real estate, not a utility bill.
Ask who owns the microgrid, because compliance follows the meter.
Follow the transformer, because supply chain is sovereignty.
Sources in the first comment.
If AI owns its own power, who owns AI?

15/04/2026

If you’re buying a device this year, read this first.

The reel covers why. This covers what to do about it.

BRAND-BY-BRAND 2026 FORECAST:
Samsung Galaxy: expect 13% phone price increase. Galaxy S series likely holds specs, but A-series (budget) will take the hit. Watch for 4GB RAM returning at entry level.
Apple iPhone: similar 13% increase. Apple has more margin to absorb, so specs may hold longer, but storage tiers will quietly shrink. The 128GB base model is under pressure.
Budget Android (Xiaomi, Oppo, Realme): Hardest hit. These brands operate on thin margins. Expect 8GB models rebranded as 6GB, 128GB storage dropping to 64GB at entry price points.
Laptops: 17% average price increase per Gartner. The real damage is spec downgrades. Last year’s $600 laptop had 16GB RAM. This year’s $600 laptop: 8GB. Same shell. Less inside.
Gaming: GPU memory costs are spiking alongside DRAM/NAND. Console refreshes may be delayed or repriced. PC gaming builds will cost 20-30% more for equivalent specs.

BUYING CHECKLIST (screenshot this):
1. Compare RAM to last year’s exact same model. If 16GB became 8GB, that’s memflation.
2. Check base storage. 256GB dropping to 128GB at the same price is the new normal.
3. Google the SSD brand inside the device. Budget laptops are swapping Samsung/SK Hynix SSDs for no-name alternatives.
4. Calculate price per GB. $600 for 8GB RAM = $75/GB. Last year’s $600 for 16GB = $37.50/GB. You’re paying double per gig.
5. Look at certified refurbished 2024/2025 models. Pre-memflation components. Often better specs than new 2026 models at the same price.
6. If your current device works: Gartner says prices stay high until 2028. Wait if you can.

WHAT THE REEL DIDN’T MENTION:
Gartner projects global PC shipments to FALL 10.4% and smartphone shipments to fall 8.4% in 2026. People are already buying less. Goldman Sachs downgraded Best Buy on April 14 citing memflation. OpenAI signed a deal with Samsung and SK Hynix for 900,000 memory wafers per month for the Stargate project. That’s where your memory is going.

Sources: Gartner Feb/April 2026, Micron Dec 2025, DOJ 2002, TrendForce Q1 2026, Goldman Sachs, Consumer Reports.

13/04/2026

The cost curve does not care about your title.
MIT tested 11,000 real tasks against 40+ AI models. The actual workers judged the results. 73% accuracy on routine work. 55% on creative tasks. The AI is ready. The economics are not. Yet.
Here is what the reel did not cover:
The study is from MIT’s FutureTech research group. They did not test hypothetical jobs. They broke real roles into granular tasks and had the people who do those tasks daily evaluate the AI output. That distinction matters because it means the accuracy numbers come from practitioners, not researchers with a benchmark.
The 86% figure comes from mapping exposed tasks back to occupational demographics. The roles most exposed (administrative support, scheduling, data entry, claims processing, coordination) are disproportionately held by women across nearly every economy studied.
The cost curve: computing costs have followed a consistent halving pattern for decades. When that curve crosses the salary line depends on local wages. North America, Western Europe, and Australia face the shortest timeline. East Asia, Eastern Europe, and parts of Latin America come next. South Asia, Southeast Asia, and Africa have the most runway, but the direction is identical.
A friend of mine said something last week I have not stopped thinking about. She said, “I am not worried about AI taking my job. I am worried that nobody will tell me when it is about to.” That line did not make it into the reel. But it is the reason I made it.
If this changed how you think about your next few years, share it with someone who needs to hear it.

09/04/2026

The startup is Doctronic. New York-based. Founded by Dr. Adam Oskowitz and Matt Pavelle. They just closed a $40M Series B led by Abstract and Lightspeed. Total funding above $65M. Over 300,000 unique weekly visitors.
The pilot launched quietly in December 2025 under Utah’s AI regulatory sandbox, a legal framework the state legislature created in 2024 that allows specific laws to be temporarily waived while regulators monitor for harm. The Department of Commerce oversees it, not the Department of Health. That distinction matters.
The 190 medications include: blood pressure, diabetes, cholesterol, birth control, SSRIs, asthma, COPD, anxiety, migraine, muscle relaxants, and erectile dysfunction. Full formulary is public on Doctronic’s site.
What is excluded: all painkillers, all ADHD medication, all controlled substances, all injectables, and anything requiring regular lab monitoring. The AI cannot start new prescriptions or modify existing ones. Refills only.

The 250-patient threshold. Before the AI operates independently in any drug class, human physicians must review its first 250 decisions in that class. After that, it prescribes autonomously. Four months in, that threshold has likely been crossed for several drug categories.
States watching: Doctronic is in active talks with Texas, Arizona, and Missouri. Their co-founder has publicly said he expects a dozen states to approve something similar in 2026.
The federal angle: H.R.238, the Healthy Technology Act of 2025, introduced by Rep. David Schweikert (R-AZ), would classify AI as a “practitioner licensed by law” eligible to prescribe any FDA-approved drug, pending state authorization. It is sitting in the House Committee on Energy and Commerce. Utah did not wait for it to pass. They moved first.
The cost of inaction: 125,000 deaths per year in the US from medication noncompliance (PMC, Duke Health, multiple peer-reviewed sources). Nearly 200,000 in the EU (European Commission, OECD). Globally, the WHO estimates half of all chronic patients do not take their medication as prescribed.

08/04/2026

The reel covers what happened. Here’s what it doesn’t.

The deal is called the Agreement on Electronic Commerce. It was led by Australia, Japan, and Singapore and took five years to negotiate. 66 WTO members signed on at MC14 in Cameroon.

Who’s in: the EU, UK, Canada, China, South Korea, Japan, Australia, Singapore, and most major economies. China being inside the deal while the US is outside is one of the least-discussed parts of this story.

Who’s out: the United States withdrew from the negotiations in 2023 under political pressure to regulate Big Tech. India, Brazil, Indonesia, South Africa, Turkey, Colombia, and Bangladesh all refused to sign. Their combined digital populations represent over 2 billion people.

India’s finance case: an estimated $500 million per year in customs duties lost under the old moratorium. Across all developing nations, the figure is roughly $56 billion in potential revenue from digital products that crossed borders untaxed.

The agreement enters into force once 45 of the 66 members complete domestic acceptance. Until then, the signatories are implementing it on an interim basis. The WTO General Council will revisit the moratorium question in Geneva, possibly May 2026.

Why AI matters here: AI models are deployed as cloud services and accessed as electronic transmissions across borders. There are currently no internationally agreed rules for how AI is classified in a trade context. Whether it’s the model itself, the API call, the data processed, or the output generated, nobody has defined which part gets taxed and by whom.

Global e-commerce is projected to hit roughly $25 trillion this year and over $83 trillion by 2035.

The rules being written right now will shape who pays what, where, and to whom for every digital product and service on the planet.

Sources: WTO, ITIF, PwC, CCIA, Business Standard.

Should countries be free to tax digital services at the border? Or does that break the system that built the internet economy?

07/04/2026

It costs roughly $9,000 to build a humanoid robot.

They sell it for $25,000. The newest model starts at $4,900.

And production just doubled in three months.

At what price does the math change for every factory, warehouse, and logistics company in the world?

We’re entering a phase where every decision has a cost.Oceans are absorbing unprecedented heat, pushing climate systems ...
24/03/2026

We’re entering a phase where every decision has a cost.

Oceans are absorbing unprecedented heat, pushing climate systems closer to their limits. Governments now face a difficult tradeoff: act aggressively and risk economic disruption, or move slowly and risk irreversible damage.

There’s no perfect path forward, only consequences.
And this isn’t a distant issue. It will shape economies, food systems, and the stability of everyday life.

What would you prioritize, immediate climate action or economic stability?
Share your perspective.

Visit Sanctity.ai to Explore more such debates

Sanctity is looking for an Algorithm Engineer to help build the core systems that determine what expertise gets surfaced...
16/03/2026

Sanctity is looking for an Algorithm Engineer to help build the core systems that determine what expertise gets surfaced, to whom, when, and why. This role focuses on designing and improving systems for ranking, matching, routing, scoring, and optimization, which are central to the platform.

You will work closely with a small, high-performing team and the founder, contributing to critical technical decisions and building systems that operate on real-world signals and feedback loops. This is a high-ownership role where your algorithmic choices will directly shape the product’s intelligence and performance.

If that sounds like something you’ll be interested in, visit our website and apply!!

Adres

Amsterdam

Website

Meldingen

Wees de eerste die het weet en laat ons u een e-mail sturen wanneer Sanctity.AI nieuws en promoties plaatst. Uw e-mailadres wordt niet voor andere doeleinden gebruikt en u kunt zich op elk gewenst moment afmelden.

Snelkoppelingen

Delen

Type