Jan 27, 2026

AI News Roundup: Anthropic Doubles Fundraise to $20B as Gemini 3 Flash, Kimi K2.5, and OpenAI Prism All Launch on the Same Day

January 27, 2026 is one of those days that makes you wonder if AI companies are coordinating their news cycles. Anthropic doubled its fundraising target to $20 billion. Google launched Gemini 3 Flash - which somehow outperforms its own Pro model. Moonshot AI dropped a 1-trillion-parameter open-source model. OpenAI launched a brand-new product for scientists. And the EU opened proceedings against Google over AI competition.

All of this lands as Big Tech earnings season kicks off with $470 billion in AI spending under the microscope.

Here’s everything that matters from January 27, 2026.

The Big Story: Anthropic Raises the Stakes to $20 Billion

Anthropic has doubled its fundraising target from $10 billion to $20 billion at a $350 billion valuation, according to the Financial Times. The round is led by GIC and Coatue, with Sequoia Capital also participating. Microsoft and NVIDIA may contribute additional capital on top of their previously announced investments.

The numbers tell a story of explosive growth: Anthropic generated close to $10 billion in revenue in 2025 and was last valued at $183 billion in September - meaning the company’s valuation has nearly doubled in four months. That $350 billion figure would make Anthropic one of the most valuable private companies in history.

This isn’t just about one company’s fundraise. It signals that institutional investors see the AI race as far from settled, even as competitors release increasingly capable models. With Google, OpenAI, and Moonshot AI all making major moves today alone, Anthropic clearly believes it needs a massive war chest to stay competitive. The question is whether $20 billion is enough - or whether this is just the opening bid in what’s shaping up to be the most capital-intensive technology race in history.

Sources: CNBC, Financial Times via MarketScreener

Today’s Top Stories

Google’s Gemini 3 Flash Outperforms Its Own Pro Model

In a move that caught the industry off guard, Google launched Gemini 3 Flash - a model that somehow beats its bigger sibling on the benchmark that matters most. Flash scores 78% on SWE-bench Verified compared to Pro’s 76.2%, while running 3x faster and costing less than a quarter as much. It also hits 90.4% on GPQA Diamond and 33.7% on Humanity’s Last Exam.

Google says it’s now processing over 1 trillion tokens per day on its API, and Gemini 3 Flash is already the default model in the Gemini app. For developers and enterprises, the message is clear: frontier-level reasoning is now available at commodity prices.

Sources: Google Blog, Google Developers Blog

Moonshot AI Open-Sources Kimi K2.5 with 1 Trillion Parameters

China’s Moonshot AI released Kimi K2.5 as open-source, a mixture-of-experts model with 1 trillion total parameters (32 billion active) trained on 15 trillion tokens. It scores 76.8% on SWE-Bench Verified and 85% on LiveCodeBench v6.

The standout feature is its “agent swarm” architecture, enabling up to 100 sub-agents executing 1,500 parallel tool calls - reducing complex task execution time by 4.5x. Bloomberg reports the release is explicitly timed to get ahead of an anticipated DeepSeek V4 launch. The open-source AI arms race in China is accelerating fast.

Sources: SiliconANGLE, Bloomberg, Hugging Face

OpenAI Launches Prism: AI for Scientific Research

OpenAI launched Prism, a free AI-native workspace designed to help scientists write and collaborate on research papers. Built on GPT-5.2 and the acquired LaTeX platform Crixet, Prism can draft and revise manuscripts with full-document context, reason over equations and citations, and pull in related literature from arXiv.

OpenAI VP Kevin Weil called it a bet on the next frontier: “I think 2026 will be for AI and science what 2025 was for AI and software engineering.” It’s available to all ChatGPT users immediately.

Sources: TechCrunch, Engadget

EU Opens DMA Proceedings Against Google Over AI Access

The European Commission opened two specification proceedings under the Digital Markets Act targeting Google. The first requires Google to give third-party AI providers equally effective access to Android features used by Gemini. The second mandates sharing anonymized search data with rival search engines and AI chatbot providers.

EU tech chief Henna Virkkunen stated the goal is to “keep the AI market open.” Google has three months to respond. This could reshape how AI services are distributed on mobile platforms.

Source: EU Digital Markets Act

Quick Hits

Big Tech Earnings: The S&P 500 hit an all-time high of 6,980.75 as earnings season kicks off. The four hyperscalers are expected to boost combined capex to $470B+ in 2026 - and investors want proof it’s paying off.
UPS Cuts 30,000 Jobs: UPS announced 30,000 additional layoffs and 24 building closures as it winds down Amazon deliveries, which it called “extraordinarily dilutive.” This follows 48,000 cuts in 2025.
Meta’s Secret Models: Meta is developing “Avocado” and “Mango” inside its Superintelligence Labs - an LLM and an image/video model, respectively. Avocado may go proprietary, breaking with Meta’s open-source tradition.
Intel 18A Ships: Intel Core Ultra Series 3 processors went on sale globally, the first chips built on Intel’s most advanced U.S.-manufactured process node. Over 200 PC designs, up to 50 NPU TOPS.
Agentic AI Surge: Databricks reports a 327% increase in autonomous agentic AI systems, with organizations reporting 60-80% reduction in processing time.
Enterprise AI Security: Zscaler’s ThreatLabz report found enterprise AI systems can be compromised in just 16 minutes, with critical flaws in 100% of systems analyzed.
AI Funding: Concourse raised $12M for AI finance agents (19x revenue growth). Pace raised $10M from Sequoia for insurance AI. SpotDraft secured $8M from Qualcomm for on-device legal AI.
AI in Healthcare: Mayo Clinic demonstrated AI detecting pancreatic cancer up to 3 years earlier than human radiologists at the J.P. Morgan Healthcare Conference.

What This Means

Today crystallizes the state of AI in early 2026: the models are getting dramatically cheaper and faster (Gemini 3 Flash beating Pro at 1/4 the cost), the open-source ecosystem is closing the gap with proprietary systems (Kimi K2.5), and the companies building these systems need unprecedented amounts of capital to keep going (Anthropic’s $20B raise). Meanwhile, regulators are moving faster than expected - the EU’s DMA proceedings against Google could force open access to AI capabilities on mobile platforms, fundamentally changing how AI reaches consumers. The $470 billion capex question looming over Big Tech earnings this week will determine whether Wall Street sees all this spending as visionary or reckless. Either way, the velocity of AI development isn’t slowing down - it’s compressing what used to take months into single days like this one.