Good morning. It's Monday, April 27, and we're covering DeepSeek’s open-source AI push, Intel’s revenue surge, and Google’s massive Anthropic bet, and more.

We're launching something new today: free 1:1 AI office hours for our community with someone who's spent years getting AI to actually work inside companies. 25 spots, first come, first served. (More details below.)

Astronomy AI Boom Strains GPUs Amid Data Explosion
Astronomers are accelerating demand for GPUs as next-generation telescopes flood the field with unprecedented data volumes. NASA’s Roman telescope will generate 20,000 terabytes over its lifetime, while the Rubin Observatory is set to capture 20 terabytes nightly, pushing researchers toward AI-driven analysis. Scientists like Brant Robertson are upgrading models like Morpheus with transformer architectures to process larger datasets faster.

Google Expands Rival Bet With Anthropic Investment
Google commits up to $40 billion to Anthropic even as it competes with its own Gemini models, underscoring a hedge in the AI race. The deal includes $10 billion upfront at a $380 billion valuation, with another $30 billion tied to performance milestones. Anthropic recently secured 5 gigawatts of compute capacity through partners including Broadcom, with capacity coming online next year. The agreement builds on Google’s earlier $3 billion-plus investment.

U.S. Flags AI Model Distillation As Security Threat
Foreign actors are reportedly copying U.S. AI models at scale rather than building their own. A White House memo warns that entities, largely tied to China, use proxy accounts and jailbreaks to extract proprietary capabilities. These distilled models mimic benchmark performance at lower cost while bypassing safeguards. The memo calls for industry coordination and accountability measures to counter industrial-scale distillation threats.

DeepSeek V4 Slashes Costs, Nears Frontier Performance
DeepSeek’s latest model delivers near-frontier AI performance at roughly one-sixth the cost of leading closed systems. The new 1.6T-parameter V4 model rivals GPT-5.5 and Claude Opus 4.7 on select benchmarks while undercutting pricing, with API costs around $5.22 per million tokens versus $30–$35 for rivals. It introduces a 1M-token context window and efficiency gains that cut compute needs to 10% of prior levels.

Salesforce has designed a four-layer architecture for companies that want to operate at agentic scale. Slack handles human-agent collaboration. Agentforce builds and deploys agents across every channel. Customer 360 brings decades of business logic to bear. And Data 360, Informatica, and MuleSoft ground every agent in unified, real-time context. It's not a patchwork of AI tools — it's a coherent system built to drive real business impact.

The agentic enterprise isn't coming. It's here. Explore Agentforce.

Intel reported April 23, 2026 quarterly revenue of $13.6 billion, up 7 percent year over year and more than $1 billion above Wall Street expectations, driven by rising demand tied to artificial intelligence. Growth was led by its data center business, where revenue climbed 22 percent to $5.1 billion as AI inferencing workloads gained traction. The company’s stock, already up over 80 percent in 2026, surged nearly 20 percent following the results.

Despite the strong revenue, Intel posted a $3.7 million loss as it continues heavy investment in manufacturing and expansion, including federally supported facilities in Arizona. The results signal momentum in Intel’s turnaround, with improving chip supply and early gains in its foundry business as customers seek alternatives to constrained competitors. → Read the full article here. (Paywall)

Google has introduced a new lineup of AI-powered smart glasses built on its Android XR platform, featuring three models tailored to different users, from casual consumers to developers. Powered by the Gemini 2.5 Pro AI system and Project Astra vision technology, the glasses enable real-time object recognition, contextual memory, and hands-free interaction.

Unlike Meta’s social-focused devices, Google’s approach emphasizes productivity, integrating tools like navigation, notifications, and workplace applications. The company is also prioritizing privacy with features such as LED indicators for active sensors and reduced sound leakage. The release positions Google to compete more directly in the growing wearable tech market, particularly among professionals and enterprise users. → Read the full article here.

Salesforce's four-layer architecture gives companies everything they need to operate at agentic scale — Slack for human-agent collaboration, Agentforce for deployment, Customer 360 for business logic, and Data 360 plus MuleSoft for unified context. Not a patchwork. A coherent system.

The agentic enterprise is here. Explore Agentforce.

AI Wolf Hoax Misleads Police: South Korean police arrested a 40-year-old man after his fake AI wolf image disrupted the Neukgu search, exposing risks of misinformation during active emergency responses.

Meta Taps Amazon Graviton Chips: Meta signs deal for millions of AWS ARM-based CPUs to power AI workloads, signaling shift toward inference and agent tasks while deepening reliance on Amazon over rival cloud providers.

Nothing Launches AI Dictation Tool: The company unveils Essential Voice, a system-level feature that converts speech into formatted text across apps with shortcuts and translation, as competition intensifies in voice-first input.

Hermes Agent Rolls Major Update: Nous Research releases v0.11.0 with 700+ pull requests from ~200 contributors, adding a new TUI, expanded LLM and image providers, and scalable multi-agent capabilities.

Microsoft Expands Copilot Agent Mode: The company rolls out “vibe working” across Word, Excel, and PowerPoint, enabling AI to generate and edit documents end-to-end as it deepens integration in Microsoft 365.

Hiten Shah has spent the last several years doing one thing: getting AI to actually work inside companies. Most recently at Dropbox, leading AI products. Before that, two decades building and advising: Crazy Egg, KISSmetrics, Nira and 1,000+ startups.

He knows what it takes to get AI working inside a real company.

We've partnered with Hiten to open up free 1:1 office hours for our community. If you're building an AI roadmap, pushing for real adoption, or trying to upskill your team and actually make it stick, this is your shot to get into it directly with him.

This won't be a sales pitch and there is no agenda (except yours!), just an honest conversation about where you are and where you're trying to go.

We're opening 25 spots. First come, first served.

→ Book your free session

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team