The IMF just raised its global growth forecast to 3.3%. It also explained why that number depends almost entirely on AI investment continuing, and what happens if the productivity gains never arrive. The fund's own model shows the upgrade could vanish twice over.
While OpenAI and Anthropic chase AGI, a wave of specialized AI tools hit the market solving problems the big labs ignored. Ten standouts from late 2025 reveal where the real value is being built: integrations, workflows, and trust.
Musk sues OpenAI for $134 billion on a $38 million donation. Trial starts April 27. Meanwhile, a solo developer built the AI memory layer that Big Tech keeps promising but never ships. Control vs. clarity. Both bets land this spring.
Musk sues OpenAI for $134 billion on a $38 million donation. Trial starts April 27. Meanwhile, a solo developer built the AI memory layer that Big Tech keeps promising but never ships. Control vs. clarity. Both bets land this spring.
San Francisco | January 19, 2026
Elon Musk's OpenAI lawsuit goes to trial in April. The ask: $134 billion for a $38 million donation. That's a 3,500x return on what was supposed to be charity. The real prize isn't cash. It's the internal documents, the Brockman journal entries, the receipts that show who promised what to whom. Discovery will be brutal for everyone involved.
Meanwhile, a former Spotify engineer named Joshua Pham spent three years building what the giants haven't: a memory layer that actually works. Enzyme connects Claude to your Obsidian vault. No VC. No team. Just one person solving the problem that Apple and Microsoft keep punting on.
The lawsuit is about control. The tool is about clarity. Both bets land this spring.
Stay curious,
Marcus Schuler
Elon Musk is suing OpenAI and Microsoft for up to $134 billion. His original contribution: $38 million between 2016 and 2020. The math works out to a 3,500x return on a charitable donation.
Trial begins April 27 in Oakland federal court. Judge Yvonne Gonzalez Rogers rejected dismissal motions from both defendants. Microsoft faces "aiding and abetting" claims that could cost them $13-25 billion if Musk prevails.
The real action is discovery. Greg Brockman's 2017 journal entries will go public. One line already leaked: "Cannot say that we are committed to the non-profit. Don't want to say that we're committed." OpenAI published a dedicated webpage cataloging what it calls Musk's "harassment." Neither side will emerge clean.
Musk's lawyers argue he was promised OpenAI would stay nonprofit. OpenAI says he demanded majority control and they refused. Both claims have documentation. Both will be tested under oath.
Why This Matters:
✅ Reality Check
What's confirmed: Trial starts April 27 in Oakland. Judge rejected dismissal. Brockman journal entries exist and will be introduced as evidence.
What's implied (not proven): Musk was specifically promised nonprofit permanence in exchange for his contributions, rather than simply assuming it.
What could go wrong: A Musk win could trigger regulatory scrutiny of every nonprofit-to-commercial AI transition, freezing deal structures industry-wide.
What to watch next: Pre-trial motions through March, and whether OpenAI attempts settlement before discovery goes public.
Joshua Pham spent three years building what Apple and Microsoft keep promising but never ship: a memory layer that actually works. His tool, Enzyme, connects Claude to your Obsidian vault. He has no investors, no team, no office. Just working code.
Pham left Spotify to solve a problem he kept running into: AI assistants are brilliant but forgetful. Every conversation starts from zero. Enzyme fixes this by indexing your notes, tags, and folder structures, then feeding that context to Claude when you ask questions.
The technical trick is compression. Enzyme achieves 2-12x token reduction versus keyword search. Instead of dumping your entire vault into the context window, it figures out where to look. "Most important knowledge is directional," Pham says. "You need a memory that tells Claude where to look, as opposed to where the answers are."
The timing is deliberate. Local-only AI inference is getting cheap enough for consumer hardware. Pham built a version running on Qwen 7B that never sends data to the cloud. He hasn't released it yet, but the option exists.
Why This Matters:
Prompt: A high-fashion editorial portrait of a woman sitting on the floor in a relaxed yet confident pose, legs folded asymmetrically, arms resting casually. She wears a black minimalist dress, vivid green tights, and glossy red high-heeled shoes. Natural makeup with bold red lipstick, short brown hair with soft bangs. The background is a solid deep red studio backdrop. Clean composition, strong color contrast between red, green, and black, retro-modern aesthetic, editorial fashion photography, soft but directional studio lighting, sharp focus, realistic skin texture, cinematic color grading, magazine-quality styling.
$2.52 trillion — Global AI spending projected for 2026, up 44% from last year's $1.65 trillion. Gartner released the forecast this week, just as Davos convenes to discuss AI governance. The money is moving faster than the rules.
Source: Gartner
Workflow of the Day: "Catch earnings call red flags before the stock moves"
Who: Investor, analyst, or portfolio manager tracking 8-12 companies per quarter.
Problem: Listening to every earnings call takes 40+ hours. You miss the tells buried in management tone shifts.
Workflow (with Claude + earnings transcripts):
Payoff: Catch management anxiety 1-2 quarters before the miss. Research time drops from 40 hours to 4.
Gotcha: Claude may over-flag normal hedging. Calibrate by comparing to competitors in the same quarter.
Tools: Claude | Seeking Alpha
Good decisions need more than pros and cons. They need the questions you forgot to ask.
The Regret Minimizer
"I'm deciding between [Option A] and [Option B]. Fast forward 2 years. For each option, describe the scenario where I most regret choosing it. Be specific: what happened, what did I miss, what was I overconfident about? Now, which regret is more recoverable?"
Best on: Claude (scenario depth and reasoning) or ChatGPT (good at generating plausible futures)
The Stakeholder X-Ray
"This decision affects [list stakeholders]. For each one, tell me: what they'll say they want, what they actually want, what they're afraid of, and what would make them a champion instead of a critic. Don't assume alignment."
Best on: Claude (psychological nuance and stakeholder modeling)
The Reversal Test
"I'm leaning toward [decision]. Argue the opposite. Not a weak strawman, your strongest case for the other path. Include one piece of evidence I'm probably underweighting and one risk I'm probably rationalizing away."
Best on: Claude (most willing to genuinely argue against your position) or ChatGPT (good at adversarial framing)
Decisions fail in the space between what you analyzed and what you assumed.
How to Generate Consistent AI Images with Ideogram Canvas
Ideogram Canvas lets you create and edit AI images with precise text rendering and style consistency. Unlike other generators, it handles typography reliably and lets you iterate on specific regions.
Tutorial:
URL: https://ideogram.ai
Anthropic is targeting a fundraise exceeding $25 billion, with Sequoia joining GIC and Coatue as major backers. GIC and Coatue are each committing $1.5 billion. This marks Sequoia's first investment in Anthropic following its recent leadership shakeup.
OpenAI's annualized revenue jumped from $2 billion in 2023 to over $20 billion in 2025, CFO Sarah Friar revealed. The company's compute infrastructure grew at a similar pace, expanding from 0.2 gigawatts to 1.9 gigawatts over the same period.
Anthropic's Claude more than doubled its web audience in December 2025, driven by developers who spent their holiday breaks intensively using the service. Users are comparing the "Claude Bender" phenomenon to the transformative moment when ChatGPT first launched.
Taiwan Semiconductor is pursuing significant expansion in the United States as Taiwan commits to spending more than $250 billion in the US as part of a trade agreement. The move reflects the complex interplay between commercial opportunity and geopolitical strategy.
The ongoing memory shortage is now rippling through GPUs, high-capacity SSDs, and hard drives. Micron says the unprecedented shortage will persist beyond 2026. Chinese smartphone makers Xiaomi, Oppo, and Vivo have already slashed 2026 shipment targets by up to 20%.
Political appointees at CISA intervened to prevent Acting Director Madhu Gottumukkala from removing the agency's Chief Information Officer. The internal power struggle raises questions about leadership stability at one of the nation's key cybersecurity agencies.
Chinese robotics firm UBTech announced supply agreements for its Walker S2 humanoid robots with Airbus and Texas Instruments, sending shares surging. The company targets production of over 10,000 humanoid units in 2026.
RunPod, an AI hosting platform founded in 2021, has reached $120 million annual revenue run rate after raising a $20 million seed round in May 2024. The company started with a Reddit post and now serves surging demand for AI infrastructure.
RISC-V chip IP company SiFive announced a partnership with NVIDIA to bring NVLink Fusion to RISC-V processors. The deal marks a major milestone for the open architecture's adoption in high-performance computing.
Periodic Labs treats nature as a training signal. The company builds autonomous laboratories where AI proposes hypotheses, robots run experiments, and results feed back into model training. 🔬
Founders
Liam Fedus led research at OpenAI connected to ChatGPT. Ekin Dogus Cubuk ran materials and chemistry efforts at Google Brain and DeepMind, including GNoME. They launched Periodic in 2025 with a critique: text-only AI hits a ceiling because the internet lacks negative results and fresh measurements. Real science requires action.
Product
Autonomous labs that close the loop. AI proposes experiments. Robots execute them. Results train better models that propose better experiments. The company started with superconductors and physical-science targets where experiments yield high signal-to-noise. Each failed experiment teaches the model what not to do, creating proprietary datasets that don't exist elsewhere.
Competition
Small but growing. Future House (AI-for-science nonprofit), Tetsuwan Scientific, materials informatics vendors, and internal R&D teams at chemical giants. The differentiator is execution: can Periodic run labs safely and reliably while producing proprietary data faster than rivals?
Financing 💰
$300M seed in September 2025. Andreessen Horowitz, DST, Nvidia, and Accel backed. Jeff Bezos and Eric Schmidt participated. The unusually large seed reflects cost: autonomous labs need hardware, robotics, lab space, safety protocols, and staff scientists. No valuation disclosed.
Future ⭐⭐⭐⭐
Periodic has pedigree, capital, and a clear philosophy. The biggest risk: scientific breakthroughs take time. Even robots can't force nature to cooperate. The witty twist: the company fights a classic AI problem. It needs new data, not new words. If the robots keep running, the models keep learning. Success may depend less on AI scientists and more on keeping instruments calibrated. 🧪
San Francisco | January 16, 2026 The story at Thinking Machines was never about ethics. It was about a company that
San Francisco | January 15, 2026 Barret Zoph told Mira Murati he was going back to OpenAI. Her response: fire him
The Trump administration approved exports of Nvidia's H200 processors to China on Tuesday. Hours later, Beijing told its tech companies to stop buying.
San Francisco | January 13, 2026 A student trains an AI exclusively on nineteenth-century London texts, asks it about 1834, and
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.
