The week of March 1st was dominated by one story: OpenAI and the Pentagon. But while that drama unfolded, Anthropic was quietly making some of the most significant enterprise moves in its history. Let's break it all down.
1. OpenAI Signs Pentagon Deal, Then Immediately Regrets It
OpenAI announced a partnership with the US Department of Defense late last week, and within 24 hours, the backlash was fierce. Critics inside and outside the company questioned whether OpenAI had abandoned its original mission. A senior hardware executive resigned over the deal.
By Monday March 3rd, CEO Sam Altman was walking it back, admitting the announcement "looked opportunistic and sloppy" and that they "shouldn't have rushed." OpenAI amended the deal to explicitly prohibit using its systems to surveil American citizens.
The whole episode revealed something important: OpenAI is no longer a scrappy research lab. It's a geopolitical actor, and its leadership is still figuring out what that means.
š” Our Take: The Pentagon deal wasn't wrong per se, it was the optics and the rush. You can't lead on AI safety and then sign government contracts without a clear policy framework. Altman knows this now.
2. Anthropic Refuses the Same Pentagon Deal, Then Gets Blacklisted
The New York Times reported this week that Anthropic and the Pentagon had been close to their own AI partnership, but talks fell apart over what Anthropic would and wouldn't allow. Dario Amodei reportedly refused to allow Claude to be used without specific ethical guardrails. The Pentagon walked away. Anthropic was informally blacklisted from some government contracts.
Anthropic is now reportedly suing over the blacklist. It's an extraordinary situation: one AI company rushes into a government deal, another refuses one, and both face significant backlash.
š” Our Take: This is AI's tobacco moment, companies are being forced to publicly declare what they will and won't do with their technology. That clarity is actually good, even if the drama is messy.
3. Anthropic Updates Claude Cowork, Serious Enterprise Plugins
On February 24th, Anthropic launched major updates to Claude Cowork, their workplace AI product. The update introduces connectors and plugins for finance, engineering, and design workflows. Finance teams can connect live market data. Engineering teams can connect GitHub, run code review, and generate documentation. Design teams can connect Figma and generate design specs from prompts.
This is Anthropic's most direct challenge yet to legacy SaaS products. When Claude can do what your $50/seat tool does, the conversation changes fast.
š” Our Take: Anthropic is executing. While OpenAI was managing Pentagon PR, Anthropic shipped the product update that actually threatens enterprise SaaS at scale.
4. Claude Sonnet 4.6, The New Everyday Model
Anthropic released Claude Sonnet 4.6 on February 17th, now shipping widely this week. It's positioned as the best balance of capability and cost: better than Sonnet 4.5 across coding, computer use, design, and knowledge work, but priced at the mid-tier API rate.
For most builders who were using GPT-4o as their default API model, Sonnet 4.6 is worth a test. The computer use improvements in particular are significant for anyone building agent workflows.
š” Our Take: Sonnet 4.6 is the model you'll actually use every day. Opus 4.6 is for when you need maximum reasoning. Know the difference and use each accordingly.
5. The Vibe Coding Mainstream Moment
The New York Times published a piece this week titled "Can A.I. Already Do Your Job?", featuring vibe coding tools like Claude Code front and center. When the Times is covering vibe coding for a general audience, it's no longer a niche trend. It's going mainstream.
The NYT piece highlighted both the opportunity (anyone can build now) and the anxiety (what does this mean for professional developers?). The answer, for now: developers who embrace AI tools are dramatically more productive. Those who don't are falling behind.
š” Our Take: The mainstream vibe coding moment is happening right now. If you're reading this blog, you're ahead of the curve. Use that advantage.
That's the week in AI. Pentagon drama, enterprise AI battles, and vibe coding going mainstream. See you next Saturday. ā”
Discover apps built by vibe coders: TheVibeBook ā
