Microsoft’s OpenAI Cloud Equation

Hello!

A quick scan across today’s AI landscape shows incumbents fighting to hold their edge while nimble specialists press forward. Microsoft is racing to lock-in OpenAI capacity just as its star partner diversifies cloud suppliers; meanwhile, Freshworks’ upbeat forecast underscores how even mid-cap SaaS players can ride the agentic wave. In Asia, South Korea’s Nota AI pushed ahead with a fast-track KOSDAQ listing, India became a magnet for WEKA’s new R&D hub, and Australia’s CBA triggered a labour row by swapping call-centre staff for speech bots. On the frontier, Amazon-backed Skild AI claims it can give every robot a shared “brain,” and water-hungry data-centre builds in Texas show the hidden environmental bill behind the boom. Down-stream, Google keeps adding consumer sparkle with its new “Canvas” study boards, and Datavault AI taps IBM watsonx to monetise corporate data. Investment heat remains blistering: Groq is circling $600 million at a reported $6 billion valuation, while Anthropic eyes an eye-watering $170 billion. Even the meme-coin crowd is getting in on the act—Google’s own models are cheekily tipping 30-fold returns for TOKEN6900. Buckle up.

Sliced just for you:

• 🏢 Microsoft’s OpenAI Cloud Equation

• 🤖 Skild AI’s “Shared Brain” for Robots

• 📈 Freshworks Hikes Guidance on AI Tailwinds

• 🛰️ Nota AI Wins Fast-Track IPO Approval

• 🏗️ CBA Job Cuts Spark Union Backlash Over Chatbots

• 🔬 WEKA Plants a Bengaluru R&D Flag

🏢 Microsoft’s OpenAI Cloud Equation

OpenAI’s expanded deals with Google Cloud, Oracle and CoreWeave have investors questioning whether Microsoft’s exclusive edge is slipping. Ahead of earnings, analysts still expect Azure’s AI-fuelled revenue to jump 35 per cent, but renegotiation of Microsoft’s equity stake and access to future models looms large. SoftBank’s proposed $40 billion OpenAI round reportedly hinges on converting the start-up into a public-benefit corporation—something Microsoft can veto. Capex, already north of $80 billion last fiscal year, will stay elevated as Redmond battles both capacity constraints and Alphabet’s $10 billion spending boost. UBS notes sentiment is split: bulls trust Satya Nadella to secure favorable terms, bears warn that multi-cloud hedging dilutes any “unfair advantage.” Either way, the talks illustrate how quickly power balances can shift when a must-have model provider starts playing the field.

🤖 Skild AI’s “Shared Brain” for Robots

Pittsburgh-born Skild AI, backed by Amazon and SoftBank, has unveiled “Skild Brain,” a foundational model designed to run on everything from assembly-line arms to humanoids. Trained on simulated episodes, human-action video and real-world telemetry, the system promises adaptive navigation, object manipulation and built-in safety force limits. In demos, bots climbed stairs, recovered from shoves and sorted dishes—tasks that stump single-purpose machines. Data from every deployed robot feeds back to the cloud, creating a virtuous spiral the founders liken to a “shared brain.” Early customers include LG CNS in logistics. After last year’s $300 million Series A at $1.5 billion, Skild has poached talent from Tesla, Nvidia and Meta to scale deployment. The move highlights how general-purpose robotics is edging closer to commercial reality, banking on generative AI’s transfer-learning mojo.

📈 Freshworks Hikes Guidance on AI Tailwinds

Help-desk software maker Freshworks bumped its 2025 revenue outlook to as high as $828.9 million, crediting surging demand for its Freddy AI suite, now adopted by 5,000 paying customers. Q2 sales hit $204.7 million—beating Street estimates—while adjusted EPS nearly doubled expectations. CEO Dennis Woodside says 1,000 government entities and marquee clients like American Express are now on the platform, using generative bots to resolve tickets and automate HR onboarding. The company claims Freddy’s upsell motion is lifting average contract values and stickiness, offsetting macro-driven IT budget caution. The raise follows a broader pattern of AI-powered guidance lifts across the mid-cap SaaS cohort, signalling that value-adding, domain-specific assistants are turning hype into hard bookings faster than many forecast.

🛰️ Nota AI Wins Fast-Track IPO Approval

Seoul-based model-compression specialist Nota AI secured preliminary fast-track approval for a KOSDAQ listing just two months after filing, bolstered by a rare “A•A” tech rating. The firm, whose edge-optimisation tools shrink and speed up large models for on-device inference, plans to float in H2 2025 with Mirae Asset as lead underwriter. Management projects 72 per cent YoY revenue growth as automakers and device OEMs embed Nota’s low-latency stack. The approval underscores South Korea’s push to cultivate domestic AI champions and deepen its chip-software value chain—part of a broader Asia-Pacific mobilisation to capture downstream AI economics amid US export controls. Investors will watch valuation comps against peers like Furiosa and Rebellions as the peninsula’s AI IPO queue thickens.

🏗️ CBA Job Cuts Spark Union Backlash Over Chatbots

Commonwealth Bank of Australia confirmed 45 redundancies tied to its roll-out of an AI voice-bot that now fields inbound customer enquiries. The Finance Sector Union claims the real figure is closer to 90 and accuses the nation’s largest lender of “excluding workers from the future.” CBA counters that it is pumping AU$2 billion into tech and frontline roles while reskilling staff for data science and prompt-engineering positions. The skirmish illustrates the social friction that accompanies rapid conversational-AI adoption in highly regulated service sectors, where cost savings clash with public-trust obligations. With Canberra exploring tighter guidelines on customer-facing automation, the episode could become a test case for balancing efficiency gains against employment and service-quality concerns.

🔬 WEKA Plants a Bengaluru R&D Flag

High-performance storage vendor WEKA opened a 50,000-sq-ft Center of Excellence in Bengaluru, expanding its Indian headcount across engineering, customer success and sales. The hub will advance NeuralMesh™, a software-defined file system that claims 90 per cent GPU utilisation for “token factories” and large-scale model training. WEKA is partnering with local cloud provider Yotta to back IndiaAI’s national GPU initiative and the Shakti Cloud. CEO Liran Zvibel says India offers unmatched engineering talent and a booming AI customer base hungry for data-infrastructure that scales faster than compute demand. About 30 per cent of Fortune 50 companies reportedly run WEKA already; the new site positions the firm to tap Asia-Pacific growth and support 24/7 follow-the-sun operations.

🛠️ AI tools updates

📝 Google Search Gets “Canvas” Study Boards

Google’s experimental AI Mode inside Search is adding a Canvas side-panel that can compile study plans, reference documents and iterative follow-ups in one persistent workspace. Users enrolled in Search Labs can click “Create Canvas,” upload PDFs or class notes and refine results with conversational prompts. Lens-powered “Search Live” also arrives, letting mobile users point their camera at an object—say, a tricky geometry diagram—and chat in real time as Gemini overlays explanations. Desktop support for image- and PDF-based Q&A is rolling out, signalling Google’s intent to knit multimodal Gemini features deep into everyday search flows rather than silo them in chatbots.

📊 Datavault AI & IBM Monetise Enterprise Data

Nasdaq-listed Datavault AI announced a multi-year pact with IBM to embed watsonx.ai across its DataScore® and DataValue® agents. The tie-up aims to help clients score data quality, price datasets and even tokenise “information assets” for Web 3.0 marketplaces. IBM will supply engineering muscle while Datavault leverages Platinum-tier partner status for co-selling. The companies tout the platform as a route to turn dormant archives into revenue streams, with early pilots in finance and healthcare. McKinsey pegs generative-AI data-monetisation potential at up to $4.4 trillion annually; Datavault hopes a ready-made stack will let mid-sized firms grab a slice without building bespoke tooling.

💵 Venture Capital updates

💰 Groq Eyes $600 Million to Scale “Chat-Speed” Chips

Sources tell Bloomberg that Groq, maker of the Language Processing Unit famed for blazing-fast token throughput, is negotiating a $600 million round valuing it near $6 billion. The deal would arm Groq against Nvidia and AMD as inference workloads shift from GPUs to specialised silicon. Proceeds are earmarked for a new 3-nanometre tape-out and global developer centers. If terms hold, it will be one of 2025’s largest semiconductor raises, further proof that investors still see upside in vertical AI hardware plays despite capital-intensive roadmaps.

🦾 Anthropic Courts a Staggering $170 Billion Valuation

Iconiq Capital is reportedly leading a $5 billion round that would price Anthropic at $170 billion—up from $90 billion just six months ago. The Claude-maker is said to be bundling preferential cloud-spend commitments and enterprise credits into the raise, echoing the structure of its earlier Google and Amazon financings. While terms remain fluid, the mooted valuation rivals legacy stalwarts like IBM, stoking debate over whether general-purpose model providers can justify such market caps before turning consistent profits.

🫡 Meme of the day

⭐️ Generative AI image of the day

Before you go, check out The inside story of building NotebookLM.