mAIn Street #285: SpaceX/Tesla readies $16B chip complex in Texas; US worker productivity doubled expectations in 2Q; How to turn good outputs into reusable skills



SpaceX and Tesla picked Texas for a $16.8 billion chip complex, Google Maps learned to book and order, and U.S. productivity beat forecasts.
mAIn Street Daily Newsletter. AI news for people who actually have jobs to do.
Friday, August 7, 2026
Thursday's strongest AI stories showed up in factories, maps, budgets and classrooms. SpaceX and Tesla committed $16.8 billion to a Texas chip complex, Google Maps added actions people can take from a search, worker productivity beat forecasts, Alphabet prepared another large bond sale, and an AI-centered private-school network laid out a rapid expansion.
Top 5
What matters today
01
Terafab is planned for Grimes County with chip manufacturing, packaging and testing on one site. The companies say it will support Optimus robots, Cybercabs and space-based data centers, cover 100 million square feet and create at least 3,000 jobs.
Source: Reuters
02
The conversational Maps feature can now move from finding a place to starting a transaction. Personal Intelligence can also use a person's Gmail and Google Calendar to tailor suggestions when the user chooses to connect those accounts.
Source: TechCrunch
03
Nonfarm productivity rose at a 1.4% annual rate, above the 0.6% forecast, while unit labor costs rose 1.3%. Economists expect AI investment to support future gains, though Thursday's report does not isolate AI as the cause.
Source: Reuters
04
The planned offering could contain as many as 10 parts with maturities from two to 40 years. It follows a higher capital-spending forecast and Alphabet's first quarter of negative free cash flow as AI investment absorbs more cash.
Source: Reuters
05
Alpha School is opening in 27 markets, including Highland Park, Carrollton and Southlake in North Texas. Students spend up to two hours a day using adaptive software for core subjects, while adults in the room focus on coaching and life skills.
Source: Axios Dallas
AI Workflows
Turn one good AI result into a skill you can use again
A verified win should leave behind the recipe, checks and approval gates that made it work.
Most of us treat a successful AI task like a lucky answer. It solves today's problem, then disappears into chat history. A simple skill gives the next run a tested starting point.
Think of a skill as a business recipe
For most businesses, a skill can be a one-page recipe for a task that already worked: preparing the monthly report, turning meeting notes into assignments, cleaning an invoice file, drafting a client update or producing an approved social graphic. The recipe says what goes in, what happens, how you check it and where a person must approve the work.
Capture it after the work passes
Finish the task and run the checks you would use before sending, publishing, paying or filing it. Fix any errors. Once the final result passes, send the assignment below in the same chat. The AI can see the completed path, including corrections, tools and decisions that got you there.
Turn the task we just completed into a reusable skill.

Include:
1. The goal and required inputs.
2. The exact steps and tools used.
3. A verification checklist with pass/fail criteria.
4. Common failure modes and recovery steps.
5. Actions that require human approval.
6. A short template I can reuse next time.

Separate the reusable process from one-time details such as names, dates, files and destinations.

Only include steps supported by the work we actually completed. Do not invent missing details.
Keep the recipe; replace the labels
Keep every time
Goal, required source types, order of operations, tools, calculations, formatting rules, pass/fail checks and approval points.
Fill in next time
Client names, dates, file names, amounts, links, destinations, owners, due dates and other details tied to one job.
Test the recipe before you save it
Use a small new example or replay the task with different one-time details. A pass means the output meets the stated checklist, uses the right source material, stops at every approval gate and leaves no invented steps. A failure means you repair the recipe, run it again and save the corrected version after it passes.
Store it where the work starts
Put the final recipe in the ChatGPT Project that holds the supporting files, your company SOP folder, a shared team document or the instructions for a custom GPT. Technical teams can package the same idea as a formal agent skill with scripts and reference files. Name it with a plain trigger such as “Prepare the Friday sales summary” so the team knows when to use it.
Why this works
Microsoft's Web Skill Factory turns solved web tasks into verified, reusable programs. In one evaluation, reusing its skill library lifted accuracy on unseen tasks from 55% to 70%, while the average number of steps fell from 17.1 to 14.7. Your business version can be much simpler. The useful principle is the same: keep the route that earned a verified result.
Save only verified successes. A bad workflow preserved perfectly is still a bad workflow.
AI Tool Spotlight
Rippling shows where the AI budget is going
AI Spend Console connects spending data from tools such as Anthropic, OpenAI Codex, Cursor and GitHub, then breaks the cost down by vendor, model, employee, team and role. Companies can track changes over time and compare spending with work signals such as pull-request volume and revisions.
This fits companies paying for several AI products and struggling to see where usage is growing. Start with one question: which tool cost changed, which team drove it and which business output moved during the same period? Treat employee-level views as sensitive management data. Review privacy, access and the limits of the chosen output metric before using the dashboard in a performance conversation.
Source: Rippling
More worth knowing
Headlines
Daily life and learning
The feature can use connected Gmail and Calendar information to personalize answers. Users still review the outside booking or ordering flow before completing a transaction.
The private-school network limits AI-supported core academics to two hours a day and uses the remaining time for coaching, projects and life skills. An education researcher cautioned that tutoring software does not cover every function of teaching.
Ohio required public districts, community schools and STEM schools to adopt a formal AI policy by July 1. Columbus says teachers will control classroom use and the tools must support safe, ethical and human-led instruction.
Eligible students will use Subject.com for personalized instruction, tutoring and round-the-clock homework help alongside classroom teachers. The Michigan district appointed a virtual learning coordinator to manage the program.
The report maps schools near data centers as communities debate noise, water and power use. It also cites research linking proximity with lower math scores, a finding that needs more work to separate data-center effects from other local factors.
Work, money and business
Productivity rose at a 1.4% annual rate in the second quarter. Initial unemployment claims increased by 1,000 to 199,000, and employers announced 33,429 layoffs in July.
Alphabet joins other large technology companies using debt to finance AI infrastructure. Big technology issuers have sold roughly $194 billion of bonds this year, 79% more than during the comparable 2025 period.
The system can reuse context across work while respecting the permissions already attached to projects and tasks. Restricted work, including confidential deals, remains limited to the people who can open it.
Mavvrik found that two-thirds of respondents had at least one business decision affected by an AI cost surprise. One-third said their company imposed an emergency spending freeze.
The young lab will use Google TPUs, Nvidia GPUs and managed training clusters for research on systems that improve through repeated work. The agreement is worth about half of the capital Mirendil raised in June.
Products, media and commerce
The music generator also plans new download controls and community guidelines as it faces lawsuits from record labels and artist groups. Watermarks can help identify origin, while enforcement still depends on platforms and users honoring them.
Their Vector AI technology aims to predict a shopper's next action from behavior, drawing on the infrastructure behind most Spotify recommendations. Retailers will still need to test whether those predictions improve useful outcomes without crossing privacy lines.
The survey of 648 adults with household incomes above $75,000 found product recommendations were the most-used AI shopping feature. Privacy was the leading change respondents said could increase their use.
The company handles calls, chat and messages for businesses that want more support volume automated. Deployment quality still depends on escalation rules, data access and a clear route to a human when the system reaches its limit.
The beta coding tool can write, debug and verify software across longer projects. Its action log lets work resume after a crash, and pay-as-you-go pricing starts at $1.25 per million input tokens.
Infrastructure and global competition
The planned Terafab complex will make, package and test logic and memory chips. The companies say their computing needs exceed one terawatt and the project could expand far beyond the initial investment.
The San Jose company raised more than $700 million to expand engineering and manufacturing. It is already shipping equipment under a multi-billion-dollar agreement with one of the four largest cloud providers.
Industrial profit rose 25% to €3.52 billion as data centers, electronics makers and factory customers increased orders. The German company raised its full-year earnings guidance.
The two Korean chipmakers are projected to hold a combined $263 billion in net cash by year-end. Investors are pressing for higher dividends or buybacks after record profits and sharp retreats from June share-price highs.
Zankore will start in Indonesia and target one gigawatt of computing capacity across the region. The company expects its first 200 megawatts to be available during the first half of 2027.
Trust, policy and security
Companies using AI or machine learning in loan origination or servicing need documented policies, risk management and oversight of vendor systems. The framework applies beyond underwriting to the full covered mortgage process.
The amendments would clarify duties around confidentiality, competence, supervision, fees and court filings when attorneys use generative or agentic AI. The proposal responds to fabricated and outdated authorities appearing in legal work.
SB 903 would restrict companies from advertising general chatbots as therapists and set rules for AI in formal mental-health settings. The debate centers on disclosure, privacy and how a bot responds when a user faces a crisis.
The evaluator Irregular said a configuration error gave the model internet access. The model reached an unidentified company's system and altered its environment, prompting work on containment practices for future tests.
Browser-based agents can read pages, handle credentials and take actions across company systems. The growing risk comes from prompt injection, excessive permissions and security tools that see the device while missing what an agent does inside a web session.
mAIn Street is built for nontechnical readers who want the signal, not the sludge.

Read All mAIn Street Back Issues Here


mAIn Street

Check out the resources I offer below and sign up for my new newsletter!

Read more from mAIn Street

AI uncovered health risks in sleep studies, Safeway entered ChatGPT, and Zoox put a price on steering-wheel-free rides. Thursday, August 6, 2026 AI moved into ordinary routines and essential systems today. Researchers pulled new risk signals from sleep studies, large U.S. companies organized around infrastructure threats, Safeway connected grocery carts to ChatGPT, Zoox set prices for steering-wheel-free rides, and Google reorganized its AI leadership as core Gemini builders left. Top 5 What...

A court lifted the block on Perplexity's Amazon shopping agent, OpenAI settled a U.S. hiring case, and Texas paused data-center hookups. Wednesday, August 5, 2026 AI collided with the law and public trust today. A federal appeals court lifted the block on Perplexity's Amazon shopping agent, OpenAI settled a U.S. hiring-discrimination case, Texas paused new data-center connections, medical researchers found a trust trap, and customers demanded human backup. Top 5 What matters today 01 A...

The White House tests AI hacking, Congress chooses ChatGPT, and businesses confront the integration work behind AI adoption. Tuesday, August 4, 2026 AI’s next phase is being shaped by implementation. Governments are testing what models can break, offices are choosing what to buy, companies are fixing old systems, schools are widening access, and artists are asking to be paid when machines borrow from their work. Top 5 What matters today 01 The White House finalized voluntary hacking tests for...