DX Today AI Daily Brief: Friday, September 11, 2026

0:00It's Friday, September 11th, 2026. You're listening to the DX Today AI Daily Brief. Today, SpaceX discloses another AI computing deal worth over a billion dollars a month. Oracle's cloud backlog tops $664 billion after a blockbuster earnings report. And co-here is reportedly in talks for a new funding round at a $20 billion valuation. Let's get into it. SpaceX has signed another blockbuster artificial intelligence hosting deal. Speaking at Goldman Sachs' Communicopia and Technology Conference on Thursday, Chief Financial Officer Brett Johnson disclosed that the rocket company will begin hosting AI Computing Workloads worth $1.11 billion a month, starting in December. A contract worth roughly $13 billion a year. The customer has not been named. Johnson said the deal supports SpaceX's target of reaching a $100 billion annual recurring revenue

1:02run rate by the end of the year. The disclosure adds to a string of AI infrastructure agreements the company has signed in recent months. As SpaceX increasingly positions its data centers and satellite network as computing infrastructure for artificial intelligence customers, not just a rocket and internet business. More big numbers from Oracle. Oracle's stock jumped after the company reported its first quarter results for its 2027 fiscal year on Thursday. Revenue came in at $19.3 billion, beating Wall Street's expectations, driven by triple-digit growth in cloud infrastructure revenue. The headline number was Oracle's remaining performance obligation, a measure of contracted but not yet recognized revenue, which reached $664 billion. Most of it tied to artificial intelligence workloads. Investors have been watching Oracle closely for signs that its enormous bet on AI data center capacity

2:05is starting to pay for itself. Even as the company continues to burn cash to build out that infrastructure. Analysts said the results suggest Oracle's AI cloud backlog is beginning to convert into real revenue. Now to funding news. Cohear is an advanced talks to raise between $2 billion and $3 billion in new funding at a $20 billion valuation according to a Bloomberg report on Friday. The terms are not final and people familiar with the discussions said the round could close as early as next week. The Toronto founded company, which focuses on selling large language models to enterprise and government customers, has been raising capital throughout the year as competition among AI labs for compute and talent intensifies. A close at the reported valuation would mark another sharp step up for Cohear. Underscoring investor appetite for enterprise-focused AI labs, even as some investors have grown more cautious about consumer facing AI startups.

3:06In acquisition news. Italian app company, Bending Spoons, has agreed to acquire online whiteboard maker, Miro. In an all cash deal valued at about $1.355 billion in enterprise value with Miro pegging the equity value near $1.79 billion, once net cash is factored in. Either way, it's a steep come down for Miro, whose valuation topped $17 billion back in 2022 at the height of the remote work boom. The deal is Bending Spoons latest acquisition of a once high flying software company at a deep discount, following its purchase of air table weeks earlier. Bending Spoons has built its business by buying struggling consumer and productivity apps and running them more efficiently. Another deal just closed. Salesforce says it has formally closed its roughly $3.6 billion acquisition of artificial intelligence. Customer service company Finn, formerly known as Intercom, completing the deal on Thursday,

4:08months ahead of the timeline the company had previously outlined. Salesforce first announced the agreement back in June and folding Finn's technology in house is meant to strengthen Salesforce's own agentic customer service products at a time when rivals are racing to automate call centers and support desks with AI agents. The early closed signals how eager Salesforce is to integrate Finn's technology and team as competition intensifies among enterprise software companies to control the AI agents handling customer interactions. Turning to new models. Chinese AI lab deep seek released a new model on Thursday called V, 4.1 Flash, describing it as the smallest model yet in its newest architecture family, but also the fastest. The company says the model adds native multi-modal capabilities, allowing it to understand images and text together along with faster throughput compared to earlier versions. Deepseek has built a reputation for releasing capable open models at a fraction of the cost

5:10of its Western rivals and the release lands the same week that several other major labs have all pushed out their own updated models. Intensifying what has become a nearly continuous cycle of AI model releases this fall. More from the coding world. AI coding company cognition released a new model called SWE2 on Thursday available immediately inside its autonomous coding agent, Devon. The company says the new model pushes what it calls the Pareto Frontier for coding agents, improving both the speed and the accuracy of Devon's ability to complete complex software engineering tasks with less human oversight. The release comes as competition among AI coding tools intensifies with several major labs and a wave of well-funded startups racing to build agents capable of handling entire engineering tasks rather than simply suggesting lines of code.

6:11Cognition has positioned Devon as a full autonomous engineer rather than a coding assistant. In product news, open AI has temporarily paused new signups and upgrades to its $200 a month, chat GPT pro plan, effective Thursday. The company said in a notice to users that existing pro subscriptions will continue to renew as normal, but it is not accepting new subscribers to the tier for now. Open AI did not give a detailed public explanation for the pause. The move follows the roll out earlier this month of Open AI's newest flagship model for professional work, which the company has been making available to pro and enterprise customers first. Open AI has previously throttled access to premium tiers during periods of unusually high demand for compute. Back to funding. Chip Interconnect startup IR Labs announced an additional $150 million in funding on Thursday, bringing its total primary capital raised this year to about $650 million.

7:12The round was backed by chip industry heavyweights, including AMD, Intel, MediaTek, Nvidia, and WeWin. IR Labs makes optical interconnect technology, sometimes called co-packaged optics, that lets AI chips communicate with light instead of electrical signals. A technology increasingly seen as necessary to link together the huge clusters of processes used to train and run the largest AI models. The company said the new funding will go towards scaling up manufacturing. More funding from Europe. French startup Arlequin AI raised 28 million euros in a series A round, announced Thursday, co-led by Redalpine and OTB Ventures. The company builds what it calls topological neural networks and has described its ambition as building a European alternative to volunteers data analysis platforms for government and defense customers. The round adds to a wave of European AI funding this year,

8:15as the continent's startups and investors argue that Europe needs its own sovereign alternatives to American AI infrastructure, particularly for government, defense, and intelligence applications where reliance on foreign technology has become a bigger political concern. One more funding round. AI security startup Enigmata emerged from stealth on Thursday with $6.5 million dollars in seed funding. The company says its technology lets AI systems learn from encrypted or otherwise sensitive data without ever exposing the underlying information. A problem that has become a bigger obstacle as companies in regulated industries like healthcare and finance look to apply AI to their most sensitive data sets. The funding round adds Enigmata to a growing list of startups focused on privacy preserving machine learning. A niche that investors have shown increasing interest in as regulators around the world, tighten rules on how companies can use personal and sensitive data to train AI systems.

9:16Finally, the workforce angle. Indian IT giant, Wipro says, artificial intelligence has freed up work capacity, equivalent to 20,000 employees over the past year, according to comments from Chief Technology Officer Sandhya Arun reported Thursday. Crucially, Wipro said those employees were redeployed to other work rather than laid off, positioning the company's AI rollout as an efficiency gain rather than a job cutting measure. The disclosure comes as companies across the technology sector face growing scrutiny over how artificial intelligence is reshaping white collar employment with some firms citing AI directly as a reason for layoffs. Wipro's framing that AI frees capacity without cutting jobs offers a notably different narrative from many of its industry peers. That's your briefing for Friday, September 11th, 2026.

10:16For DX Today, stay curious.