2026-10-27
By Vadym · Generated with AI, curated by me
Today’s stories aren’t about new models — they’re about AI’s business plumbing catching up. Anthropic put real numbers behind an IPO, a chip company turned down a buyout to raise its own money, and law firms and a pharma giant are paying real cash for agents that do real work.
Anthropic made its S-1 filing public on October 26, confirming it’s aiming for a mid-November IPO. The filing shows $16.3 billion in H1 2026 revenue ($4.73B in Q1, $11.5B in Q2) against $4.6 billion for all of 2025, alongside an $80.7 billion net loss that the company attributes mostly to non-cash charges from marking convertible financing instruments to its rising valuation. About 25% of revenue reportedly comes from just two unnamed customers.
The UK government announced on October 23 that it will fund a new AI Biosecurity Lab with £100 million in public money, with the three labs adding up to £200 million more in compute credits and model access. OpenAI is contributing roughly £100 million in compute and model credits plus access through its Rosalind Biodefense programme; Anthropic is providing comparable credits and technical support; Google DeepMind’s scientists will work directly with the lab’s biosecurity researchers.
Jim Keller’s AI chip startup Tenstorrent closed a $1.5 billion Series E at a $6.2 billion valuation, more than double its prior $2.6 billion mark, co-led by Trace Capital and Innovation X with ARK, Fidelity, Morgan Stanley and T. Rowe Price joining in. The company says it turned down acquisition offers from both Intel and Qualcomm — Qualcomm’s reportedly reaching $8–10 billion — after a May demo showed its Galaxy servers running DeepSeek models at roughly $6 per million tokens versus about $30 on Nvidia’s GB300 NVL72 rack, a company-reported comparison.
Swedish legal AI company Legora announced a $550 million Series E on October 21, led by Atreides Management with new backers including Fidelity and J.P. Morgan, pushing its valuation to $8.8 billion — up from $5.6 billion in March. The company says annual recurring revenue now tops $200 million, with more than 140,000 legal professionals across 2,400-plus firms using its “aOS” platform.
A TechRepublic tally updated October 23, citing TrueUp data, puts 2026 tech layoffs at roughly 196,000 workers, with October bringing fresh cuts at eBay, Netflix and Electronic Arts on top of earlier rounds at Amazon (~16,800), Meta (~8,700) and Microsoft (~4,800). Other trackers put the running total higher, above 225,000, with AI or automation cited in roughly 36–41% of tracked events.
Pathos AI closed a $237 million Series E on October 23, led by Bristol Myers Squibb, valuing the company at roughly $3 billion. Alongside the raise, BMS is adopting Pathos’s “Foundry” platform — which runs thousands of agents estimating a clinical trial’s probability of success — across its R&D pipeline, while Tempus AI is licensing Pathos a dataset of hundreds of thousands of neurological-disorder patients for a new Alzheimer’s-focused program.
None of today’s stories are about a new model. They’re about the business plumbing catching up to what models can already do: an IPO filing that puts real numbers on the table, a chip company betting independence beats a buyout, law firms and a pharma giant paying real money for agents that do real work, a government putting compute behind biosecurity risk, and a layoffs tally that keeps climbing instead of leveling off. The experimental phase is quietly ending in several verticals at once — what’s left to settle is who captures the value, and who absorbs the cost.
— Boba
Curated by Vadym