In partnership with
It hasn’t been a smooth start to 2026 for OpenAI. Competitive pressure from Anthropic's Claude and Google's Gemini has intensified, with Claude Opus 4.6 topping benchmark rankings and ChatGPT losing market share for the first time. Then came a roller coaster of a week.
What connects these events is a company trying to consolidate its position through financial and political force at a moment when public trust in AI companies is fragile. Each move individually made strategic sense for OpenAI. Together, they created a narrative that handed its biggest competitor the best marketing campaign money could never buy.
CEO Sam Altman also appears decreasingly like one of the good guys by the day. Full conversion into Darth Vader (as with Elon Musk) feels imminent.
What went down
Monday: OpenAI cut its 2030 spending forecast from $1.4tn to $600bn. I wouldn’t be surprised if this goes down again with 12 months.
Wednesday: Altman bizarrely compared AI energy use to "20 years" of raising a human, sparking a social media backlash. He had an absolute howler there.
Friday morning: OpenAI closed $110bn from Amazon, Nvidia and SoftBank. That’s a lot of money - indeed, it’s the largest private raise ever.
Friday evening: OpenAI signed a Pentagon deal hours after Anthropic was dropped. Cue Altman arch villain accusations.
Saturday: Claude hit number one on the App Store amid a second social media backlash. Retribution.
Why does this matter?
The Pentagon situation is the most consequential development. Anthropic was designated a "supply chain risk to national security" for refusing to allow its models to be used for mass surveillance and autonomous weapons (we covered this last week). OpenAI then secured a deal it says includes the same restrictions Anthropic was punished for requesting. The first-ever use of a supply chain risk label against an American company raises serious questions about government-business relations in AI. In my view, it’s extremely bizarre.
The consumer backlash was immediate. A Reddit post titled "Cancel and Delete ChatGPT" attracted 30,000 upvotes. Katy Perry, who always seems to pop up at the strangest moments, posted a screenshot of Claude's pricing page. Anthropic reported daily signups tripling and paying subscribers more than doubling this year. Whether this translates to lasting market shift or fades, but the whole episode has been extraordinary PR for Anthropic. Being ahead of the curve, we ditched ChatGPT for Claude last Autumn.
Partner message
Attio is the AI CRM for modern teams.
Connect your email and calendar, and Attio instantly builds your CRM. Every contact, every company, every conversation, all organized in one place.
Then Ask Attio anything:
Prep for meetings in seconds with full context from across your business
Know what’s happening across your entire pipeline instantly
Spot deals going sideways before they do
No more digging and no more data entry. Just answers.
Our take
The week exposed a tension at the centre of OpenAI's strategy. The company is simultaneously positioning itself as the industry's most commercially aggressive player, with $110 billion in fresh capital, a Pentagon contract and 900 million weekly users, while trying to maintain the trust of a user base that increasingly cares about where AI gets deployed. The Pentagon deal illustrated this perfectly: Altman publicly supported Anthropic's safety principles in the morning and announced he had taken their contract by evening. Alongside his out of touch comments about training humans in India, it was a calamitous week of PR.
The funding round itself tells a story about where AI is heading. Amazon's $50 billion investment comes bundled with AWS becoming OpenAI's exclusive third-party cloud provider for enterprise. Nvidia's $30 billion comes with commitments for OpenAI to consume gigawatts of its next-generation hardware. These are infrastructure lock-in deals dressed as investments.
Meanwhile, OpenAI quietly cut its own spending projections from $1.4 trillion to $600 billion, signalling that even the best-funded AI company is recalibrating what it can realistically build. The question is whether all that capital can solve a problem that money alone may not fix: a growing perception that OpenAI will take any deal on the table.
Another big thing… Perplexity Computer
Also on Wednesday, Perplexity AI launched Perplexity Computer, a platform that orchestrates 19 AI models into a single autonomous workflow system. Rather than answering one prompt at a time, Computer takes a high-level goal, breaks it into subtasks and assigns specialised sub-agents to handle research, coding, design and deployment in parallel.
CEO Aravind Srinivas said the team had been building it quietly for two months. The product is available to Max subscribers at $200 per month and represents Perplexity's bet that the future of AI is multi-model orchestration rather than any single system.
