The Inference Report

Archived Reports

March 12, 2026
120 feeds · 102 labs · 18 repos

The infrastructure layer is consolidating while the application layer fragments into a thousand bets on what agentic AI actually means in practice. Nvidia is spending 26 billion to build open-weight models and striking 2 billion deals with cloud providers like Nebius, essentially funding its own cus…

March 11, 2026
65 feeds · 17 labs · 500 papers · 21 repos

Today's news marks a genuine inflection point: the age of standalone AI products is ending, and the era of embedded AI as operational infrastructure is beginning. The evidence is structural, not speculative. Google, Amazon, and Zoom are not adding AI features to their platforms; they are embedding A…

March 10, 2026
52 feeds · 12 labs · 20 repos

Meanwhile, the AI industry is fracturing along a new fault line between those building for government and those refusing to. Anthropic's lawsuit against the Pentagon over its supply-chain-risk designation has drawn public support from more than 30 OpenAI and Google DeepMind employees signing an amic…

March 9, 2026
11 feeds · 104 papers · 19 repos

The market is moving faster than the institutions meant to govern it. Ring's privacy explanations aren't holding back adoption; Anthropic's Pentagon contracts aren't stopping other startups from taking defense work; Microsoft's AI agent licensing isn't waiting for safety consensus. Meanwhile, the ac…

March 8, 2026
8 feeds · 19 repos

The gap between what AI companies profess and what they actually build is no longer theoretical. Anthropic's Pro-Human Declaration preceded its Pentagon contract announcement by weeks, OpenAI's robotics team lost an engineer rather than accept defense work, and Google tied a $692M executive compensa…

March 7, 2026
45 feeds · 10 labs · 20 repos

Anthropic's refusal to surrender control of Claude to Pentagon demands has split the AI market along a fault line that now determines winners and losers: companies that preserve user trust versus those that accept government terms. The choice cost Anthropic a $200 million contract but delivered some…

March 6, 2026
46 feeds · 13 labs · 319 papers · 12 repos

The Pentagon's power over AI builders became concrete this week when Anthropic refused to grant unrestricted military access to Claude and faced immediate retaliation: a federal agency ban, a supply-chain risk designation, and isolation as the first American AI firm to receive it. Within hours, Open…

March 5, 2026
30 feeds · 5 labs · 295 papers · 22 repos

Much like the shift from open-source Unix systems to proprietary enterprise software in the 1990s, the AI market is reorganizing itself around control rather than capability. The Pentagon's swift punishment of Anthropic for refusing to lift military restrictions on Claude, followed by OpenAI's immed…

March 4, 2026
69 feeds · 8 labs · 290 papers · 35 repos

Control is consolidating at the infrastructure layer while fragmenting everywhere else. Microsoft, Meta, and Amazon are locking in chip supply and power deals that give them permanent advantage over software vendors trying to build on top. Block cut 40 percent of its workforce because AI tools made…

March 3, 2026
40 feeds · 4 labs · 499 papers · 18 repos

The AI industry's most significant transformation this week is not happening in model benchmarks or research papers but in the physical infrastructure that underpins everything. OpenAI announced a $110 billion raise at a $730 billion valuation, with $30 billion each from SoftBank and NVIDIA plus $50…

March 2, 2026
9 feeds · 180 papers · 32 repos

The AI industry's carefully constructed narratives are colliding with an uncooperative reality across multiple fronts. The Pentagon's demand that Anthropic accept military use terms by February 27 or risk designation as a supply chain risk represents the most explicit government intervention in AI c…

March 1, 2026
20 feeds · 42 labs · 38 repos

The AI industry's central tension is no longer whether agents will work, but whether anyone can afford to run them at scale. That double bind defined the week's news across every sector. The Pentagon gave Anthropic an ultimatum that made plain what has been true for months: the era of AI companies n…