In every technology cycle, strategy evolves, markets shift, and products get rebuilt, but leadership is the only true constant.
The iconic companies aren’t built by the smartest people. They’re built by the most human ones.
I’ve had the privilege of sitting down with the CEOs of NVIDIA, Microsoft, and Intel.
Different eras. Different markets. One consistent pattern: People First. Always.
The Power of Adaptability: @Jensen Huang told me about making decisions that felt “life-threatening” to NVIDIA, only to realize later they were wrong. His superpower isn’t just vision; it’s the courage to adapt. When the facts changed, he changed his mind. Adaptability compounds. Ego doesn’t.
Innovation Through Empathy: @Satya Nadella reshaped Microsoft by moving from a culture of “know-it-alls” to “learn-it-alls.” He proved that empathy is an innovation strategy. By seeking to understand the “unmet, unarticulated needs of customers,” he turned vulnerability into a competitive advantage.
Humility & Strategic Partnerships: @Lip-Bu Tan’s advice is timeless – stay humble and listen deeply. He operates on a culture of “bad news first,” flattening hierarchies to hear the truth from the trenches. In a hyper-competitive market, your credibility is your only true moat.
Here’s what I believe:
In every technology wave – semiconductors, software, Internet, cloud, mobile, AI – tools change. Leadership fundamentals don’t.
The founders who win in the long term are not the loudest. They are the ones who:
Are mission-oriented
Build a culture of teamwork and empathy
Adapt fastest, as dinosaurs never survive
Listen hardest
Focus insanely, as companies die of indigestion, not starvation
Believe that company building is a marathon, not a sprint
And put people before pride by being People First
In the AI era, especially, where speed and disruption dominate headlines, people-first leadership is not optional. It’s the moat.
Now let’s turn to this week’s signals, where the headlines may focus on capital, infrastructure, and model breakthroughs, but the enduring advantage still comes down to leadership and execution.
Signals Shaping the Future of AI:
Infrastructure
MWC 2026 showcased major AI-native networking moves from AMD, NVIDIA, Qualcomm, Intel, and others as the industry eyes 6G. Vendors highlighted new AI chips, telecom platforms, and cross-industry partnerships aimed at moving AI from pilot projects into live network deployments. Click here
Flex expands U.S. manufacturing of AMD Instinct AI platforms. Flex will produce AMD’s MI355X 8-GPU systems at its Austin facility, with volume ramp-up next quarter, supporting domestic capacity for large-scale AI data center deployments. Click here
Huawei unveiled its Atlas 950 AI SuperPoD at MWC 2026, connecting up to 8,192 Ascend chips into one large AI system. The launch signals Huawei’s push to compete directly with Nvidia and AMD in high-end AI data center infrastructure using its own chips and software. Click here
Intel unveiled its next-gen Clearwater Forest Xeon 6+ chips on its 18A process, targeting 6G and edge AI networks. The processors aim to deliver higher performance with lower power use for telecom and data center operators. Click here
Enterprise
Amazon launches Amazon Connect Health for AI-powered clinical workflows. The AWS platform handles patient calls, scheduling, ambient documentation, and automated billing, integrating with Epic as Amazon deepens its healthcare AI push. Click here
OpenAI launches GPT-5.4 with native computer-use capabilities. The new model combines improved reasoning, coding, and document handling with the ability to operate a user’s device via keyboard, mouse, and API calls, positioning it as a major step toward more autonomous AI agents. Click here
Anthropic’s Claude tops Apple’s U.S. App Store following Pentagon dispute. The chatbot climbed from outside the top 100 in January to No. 1, overtaking ChatGPT, as daily sign-ups hit record highs and paid subscriptions more than doubled this year. Click here
Meta signs a multiyear AI content licensing deal with News Corp worth up to $50 million annually. The agreement, set to run at least three years, gives Meta access to The Wall Street Journal owner’s U.S. and U.K. content for AI training and retrieval in its AI products. Click here
Capital Flows
OpenAI’s former chief research officer is reportedly raising ~$70 million for a new AI manufacturing automation startup. The company aims to apply advanced AI systems to industrial and factory workflows, according to WSJ reporting. Click here
NVIDIA to invest $4 billion in photonics firms Coherent and Lumentum. The chipmaker will deploy $2 billion into each company under multi-year strategic agreements, securing supply and advancing silicon photonics to support next-generation AI data center infrastructure. Click here
Ayar Labs raises $500 million to scale optical interconnects for AI data centers. The Series E round values the company at $3.75 billion and will fund production of its co-packaged optics technology aimed at boosting AI infrastructure performance. Click here
Anduril is reportedly targeting a $60 billion valuation in a new funding round. The defense-tech company is said to be raising billions in a round led by Thrive Capital and Andreessen Horowitz, following its $2.5 billion Series G at a $30 billion valuation last year. Click here
Netflix acquires Ben Affleck’s AI filmmaking startup InterPositive. The streaming giant bought Ben Affleck’s AI film-tech company, with the full team joining Netflix and Affleck staying on as a senior adviser. Click here
Research
Microsoft unveils Phi-4-Reasoning-Vision-15B, a compact multimodal AI model that selectively activates chain-of-thought reasoning. The 15B-parameter model processes text and images, was trained on ~200 billion tokens in four days on 240 Nvidia B200 GPUs, and is optimized for efficient, latency-sensitive enterprise workloads. Click here
Google DeepMind launches Gemini 3.1 Flash-Lite as a faster, lower-cost multimodal model. The model supports up to a 1M-token context window and is optimized for high-volume, latency-sensitive tasks like translation and classification across text, image, audio, and video. Click here
Anthropic published a new “observed exposure” metric for AI displacement risk, blending task feasibility with real Claude usage. It finds no clear unemployment spike since 2022 for highly exposed roles, but early signs of slower hiring for ages 22 to 25 in the most exposed occupations. Click here
Gartner forecasts AI spending will reach $2.5 trillion in 2026. The firm expects a 44% year-over-year increase, with cloud providers, chipmakers like AMD, and software platforms such as Datadog positioned to benefit from sustained enterprise AI investment. Click here
Policy
U.S. bans Anthropic’s Claude for federal use, citing supply-chain risk. The Pentagon directed agencies and defense contractors to phase out Anthropic’s AI tools, triggering contract reviews and compliance shifts across government suppliers. Click here
xAI loses bid to block California’s AI data transparency law. A federal judge denied Elon Musk’s company a preliminary injunction against the state’s rule requiring AI firms to disclose summaries of training data, allowing the law to remain in effect as the case proceeds. Click here
OpenAI reaches a deal to provide AI models to the Pentagon under legal-use constraints. The agreement allows classified military use while citing existing laws on autonomous weapons and surveillance, after Anthropic’s refusal to relax safeguards led to its federal ban and supply-chain risk designation. Click here
Global AI Strategy
Tech giants sign White House pledge to fund power for AI data centers. Google, Microsoft, Meta, Amazon, Oracle, xAI, and OpenAI are committed to securing or building dedicated electricity capacity and covering grid upgrade costs to prevent consumer energy price increases ahead of the U.S. midterms. Click here
China’s new five-year plan makes AI central to its economic strategy. The blueprint pushes nationwide AI adoption and targets breakthroughs in chips, quantum, 6G, and humanoid robotics to strengthen technological self-reliance. Click here
NVIDIA and global telecom leaders commit to AI-native 6G infrastructure. Partners, including Cisco, Ericsson, Nokia, Deutsche Telekom, SK Telecom, and SoftBank, will collaborate on open AI-RAN platforms to embed AI across next-generation wireless networks. Click here
UK launches £40 million frontier AI research initiative. The government announced new funding to support advanced AI research, positioning the investment as part of a broader push to strengthen national technological sovereignty and long-term competitiveness. Click here
Talent Signals
Each week, we spotlight key roles tied to the themes shaping this week’s AI headlines, connecting talent to the companies driving the news.
@TogetherAI provides infrastructure and tools that simplify building, training, and deploying generative AI models at scale. TogetherAI’s platform supports teams operating models and workflows with reliability and efficiency. Open roles are listed on its careers page. Click here
@Rogo.ai builds AI tools that generate real-time sales intelligence, automatically capturing key insights and workflows from conversations to help revenue teams close deals faster. Rogo is expanding its engineering and product teams. Open roles are listed on its careers page. Click here
@Pika builds an AI video creation platform that turns text, images, and prompts into dynamic, shareable video content without traditional editing tools. The company is hiring across engineering, research, and product roles to advance its generative video and editing models and scale creative tooling for creators and teams. Open roles are listed on its careers page. Click here
You can see all the opportunities at Mayfield-backed AI companies here, and across the broader ecosystem here.
Social Signals
The most important conversations in AI are unfolding across social media, where top voices are shaping the next wave of signals and strategy. Here are some of the top social signals and their takes from the past week.
Allie K. Miller (Click here) — “Not a single person thinks that their setup is 100% secure. If you’re not okay with all of your data being leaked onto the internet, you shouldn’t use it. It’s a black and white decision.” In a post viewed 743K+ times, Miller shares notes from a sold-out OpenClaw meetup in NYC where builders are running multi-agent systems with distinct roles, spending thousands per month on tokens, and accepting meaningful security risk in exchange for capability. She describes agents that manage other agents, build live products and presentations, and blur the line between work and play, capturing a fast-moving, experimental edge of the AI agent ecosystem.
Matt Shumer (Click here) — “I’ve been testing GPT-5.4 for the last week. In short, it is the best model in the world, by far. It’s so good that it’s the first model that makes the ‘which model should I use?’ conversation feel almost over.” In a post viewed 1.3M+ times, Shumer says even the standard version with heavy thinking outperforms prior Pro models, breaking his habit of defaulting to higher-tier options. He calls coding performance inside Codex “essentially flawless” and says Pro mode is now overkill for most use cases, while noting remaining gaps in frontend taste and real-world context awareness.
Ajeya Cotra (Click here) — “On Jan 14, I predicted that the SWE time horizon by EOY would be ~24 hours. Now I think it’ll be >100 hours, and maybe unbounded. For the first time, I don’t see solid evidence against AI R&D automation this year.” Cotra is materially revising her outlook as models rapidly extend the length of software tasks they can handle autonomously. For founders, this is not an abstract benchmark shift. If AI can own 100-hour workstreams end-to-end, the bottleneck shifts from engineering capacity to product vision, verification, and iteration speed. Small, high-agency teams begin to scale like much larger ones, and the compounding advantage goes to those who redesign workflows around that reality early.
