Key Moments
Former Intel CEO on What Went Wrong, What's Next + Lovable CEO on the Real Promise of Vibe Coding
All-In PodcastWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
Intel lost its dominance by prioritizing business over engineering, leading to missed opportunities like Apple's silicon. Meanwhile, AI's rapid advancement is limited by energy capacity but promises a two-decade buildout revolutionizing industries.
Key Insights
Intel's shift from technical leadership to business-focused management, coupled with a $100 billion return to shareholders via dividends and buybacks in the years before Pat Gelsinger's return, led to a failure to invest in new factories and critical equipment like EUV machines.
Apple's decision to develop its own silicon was a response to perceived limitations from Intel as a supplier, with Steve Jobs having already prepared core technologies internally for this transition over several previous years.
NVIDIA's success with GPUs wasn't initially planned for AI or cryptocurrency; it stemmed from building high-performance throughput machines, with the development of their CUDA software stack being a critical turning point for general-purpose computing applications.
TSMC's foundry model, focusing on manufacturing for the entire industry rather than just internal needs, allowed them to scale significantly, producing 5 times the wafers of Intel when Gelsinger returned, and becoming the industry standard.
A blockade of Taiwan could cripple the global economy within three weeks due to energy reserves, leading to fab shutdowns that take 90 days to recover from, with an economic impact potentially exceeding the Great Depression.
Lovable is seeing exponential growth, with over a million new projects built weekly on its platform, and over 700 million visits to applications monthly, demonstrating a significant shift in software development towards no-code/low-code solutions empowered by AI, with 700 million visits to applications each month and 50 million apps built on the platform to date in 20 months.
The AI buildout is currently capped by energy capacity, which is expanding at 4-5% globally, preventing an unchecked bubble, but the potential value of AI in improving supply chains, finance, and logistics is considered 'somewhat infinite'.
Quantum computing is predicted to yield meaningful results and demonstrate quantum supremacy across multiple industries by 2030, impacting fields like chemistry and biology with capabilities beyond today's computation.
Intel's strategic missteps and the rise of a technologist-led vision
Pat Gelsinger, a former Intel CEO with 34 years at the company, reflects on Intel's decline from an American tech giant to a position where it was surpassed by competitors like NVIDIA, TSMC, and Apple. He attributes a significant part of this downfall to a shift in leadership from deeply technical individuals to business and finance-focused managers. Gelsinger highlighted that during the years leading up to his return as CEO, Intel distributed $100 billion to shareholders through dividends and stock buybacks. This financial strategy, while popular, came at the expense of crucial investments in R&D and modern manufacturing infrastructure, such as building new factories or acquiring essential machinery like EUV machines. He argues that critical technological decisions, especially those involving billions of dollars, should be guided by technologists who understand the long-term implications and evolving tech trends, rather than solely by spreadsheet economics. Gelsinger's return aimed to re-center Intel around its technical roots and to advocate for long-term, risky investments that might not have immediate financial justification but are vital for future competitiveness.
Apple's independent chip design strategy
The conversation delves into Apple's decisive move to design its own silicon, a pivotal moment that shifted its relationship with Intel. When Steve Jobs became dissatisfied with Intel's ability to meet Apple's evolving demands for smaller, lower-power chips, he initiated a project to bring chip design in-house. This was not an overnight decision, but rather a gradual build-up of core competencies, leveraging acquisitions and internal development. Gelsinger recounts a key moment where Steve Jobs revealed that Apple had already been porting its operating system to the x86 Intel architecture for the past four releases, demonstrating a proactive strategy to control its destiny. This move allowed Apple to optimize system design with silicon design in tandem, a flexibility Intel, focused on a broader Windows ecosystem, couldn't easily provide.
NVIDIA's accidental triumph in AI and crypto
The discussion touches upon Jensen Huang's vision at NVIDIA, focusing on graphics cards as high-performance throughput machines. Initially, Intel and others viewed these as niche products for gaming. However, NVIDIA's commitment to building a robust software stack, particularly with technologies like CUDA and SIMT, transformed their GPUs into powerful, general-purpose computing devices. This evolution was serendipitous, as these computational capabilities proved highly applicable to cryptocurrency mining and, critically, to artificial intelligence workloads. Gelsinger noted that AI had experienced multiple 'nuclear winters' by that point, making its widespread adoption seem unlikely, yet NVIDIA's continuous improvements in hardware and software laid the groundwork for its eventual explosive growth in these fields, a path Intel explored but ultimately abandoned in a project called 'Larabe'.
TSMC's foundry model revolutionizes manufacturing
A significant factor in Intel's competitive disadvantage was the rise of TSMC and its foundry model. While Intel operated as an Integrated Device Manufacturer (IDM), designing and manufacturing its own chips, TSMC adopted a pure-play foundry approach, dedicating its manufacturing capabilities to any company's designs. This strategy, requiring immense capital investment and continuous engineering innovation, allowed TSMC to become the world's leading chip manufacturer. Gelsinger points out that Intel, at the time, did not open its manufacturing processes to third parties, viewing its internal capacity as sufficient. TSMC's unwavering focus on being the manufacturing partner for the industry, coupled with customer demand from companies like Apple, propelled its growth. By the time Gelsinger returned to Intel in 2001, TSMC was already producing five times the wafers of Intel, underscoring the success of the foundry model, which has since become the industry standard except for Intel and memory manufacturers.
Geopolitical risks and supply chain resilience
The conversation highlights the critical geopolitical risks associated with semiconductor manufacturing concentrated in Taiwan. The island has less than three weeks of energy reserves, meaning a blockade could quickly lead to widespread power outages. Importantly, semiconductor fabrication plants (fabs) require continuous operation; a shutdown for even 90 days can render them unusable without extensive recalibration. The economic impact of such a disruption would be catastrophic, potentially exceeding that of the Great Depression. Gelsinger emphasizes the urgent need for more resilient supply chains, noting that China's repeated military exercises in the Taiwan Strait suggest clear intentions. While the CHIPS Act is increasing onshoring efforts, with the US manufacturing about 18% of leading-edge chips compared to 12% previously, there's still a long way to go to achieve significant self-sufficiency.
The AI revolution and its energy constraints
The current AI buildout is described as potentially 'the buildout to end all buildouts,' requiring vast data centers and an unprecedented number of chips for both training and inference. However, Gelsinger believes that energy capacity acts as a natural brake on an unchecked AI bubble, with global energy expansion at around 4-5% and specific improvements in the US grid. He posits that without sufficient energy, the massive investment in GPUs and data centers would be unsustainable. Despite this constraint, the potential value of AI, particularly in improving efficiency across logistics, finance, and supply chains, is seen as 'somewhat infinite.' Gelsinger forecasts a buildout spanning decades, not just years, and aims to make AI 10,000 times cheaper and more efficient, potentially driving a new era of technological advancement.
Lovable: Empowering human creativity through AI development
Anton Osika, CEO of Lovable, discusses his company's mission to empower humans to build software. Lovable has experienced remarkable growth, with over a million new projects built weekly and 700 million monthly app visits, processing 50 million apps built on the platform in 20 months. This success is attributed to leveraging AI to enable both technical and non-technical users to rapidly create functional applications. Lovable provides a structured, opinionated framework that incorporates best practices for software architecture, security, and payments, removing significant technical hurdles. The platform allows users to move from conceptualization to viable product deployment in days, often at a fraction of the cost and time of traditional development methods. The company sees its platform as an 'AI co-founder,' assisting users not just in building but also in operating and strategizing for their businesses.
The future of computing: Quantum and integrated systems
The discussion turns to the future of computing, particularly quantum computing. Gelsinger predicts that meaningful quantum computing results will emerge this decade, with quantum supremacy demonstrated by 2030, impacting fields like chemistry and biology that are currently beyond computational reach. While quantum computing has been 'five years away' for decades, Gelsinger is optimistic about progress, citing advancements in qubit stability, error correction, and algorithm development. He also anticipates the convergence of classical computing, AI computing, and quantum computing, forming a 'trinity' that will unlock unprecedented capabilities. This integration is expected to drive significant innovation in solving complex problems, from logistics optimization to breaking current encryption standards.
Mentioned in This Episode
●Products
●Software & Apps
●Companies
●Organizations
●Books
●Concepts
●People Referenced
Common Questions
Intel's decline is attributed to a shift from technical leadership to business-focused management, a failure to invest in crucial technologies like EUV machines and custom silicon for products like the iPhone, and underestimating the growth of companies like NVIDIA and TSMC.
Topics
Mentioned in this video
The company behind Grok, mentioned as a maker of inference silicon.
A major American technology company that was once a leader in microprocessors, but faced decline due to strategic misses and competition from NVIDIA, TSMC, and Apple.
A technology company that initially focused on graphics cards but became dominant in AI and high-performance computing, surpassing Intel.
Taiwanese Semiconductor Manufacturing Company, a leading chip manufacturer that adopted a foundry model, becoming crucial for many tech companies and surpassing Intel in wafer production.
A technology company that transitioned to designing its own silicon (Apple Silicon), a move that significantly impacted Intel and other chip manufacturers.
A company offering an AI-native platform for global accounts, cards, and payments, designed for the intelligent era.
A company acquired by Apple to build its competency in semiconductor design.
A company developing AI hardware, mentioned in the context of impressive companies in the AI space.
A company mentioned for its significant revenue, in contrast to early dot-com era companies.
A quantum computing company and a portfolio company of the speaker, mentioned in the context of quantum computing.
A CRM and business software company. Lovable aims to interoperate with tools like Salesforce.
A CRM and marketing software company. Lovable aims to interoperate with tools like HubSpot.
Former CEO of Intel, remembered as a deeply technical leader and mentor.
Co-founder of Intel. His name was pronounced 'Gordon Moore'.
CEO of Microsoft, noted as a technically adept leader, even though not a founder.
CEO of Google and Alphabet, mentioned as a deeply technical individual.
Co-founder of Apple, known for his leadership, ruthlessness, and vision, including the decision for Apple to create its own silicon.
Author known for his science fiction novels, quoted for his observation that 'the street finds its own use for technology'.
Author of a biography on Steve Jobs, recommended by the speaker.
NVIDIA's parallel computing platform and programming model, which enabled GPUs to be used for general-purpose computing and AI.
An AI chatbot from xAI, mentioned as an example of companies making inference silicon.
A platform that aims to empower anyone to build software and businesses, gaining significant traction and revenue.
A tool that captures conversations hands-free and turns them into notes and next steps, marketed for leveraging conversations that happen outside of formal meetings.
Amazon Web Services, a cloud computing provider. Lovable works with AWS under the hood for its hosting services.
A suite of productivity software. Lovable aims to interoperate with Microsoft products.
A new AI model from Anthropic, discussed as an example of a massive step function in model capabilities, though human planning remains crucial.
A project at Intel aimed at leveraging the x86 architecture for broader computational tasks, but was canceled after the speaker's departure.
A European research organization for particle physics where the speaker was introduced to the concept of 'co-opetition'.
A large US company where 'NAD+' used Lovable to build new product lines and replace internal tools.
A pre-accelerator program mentioned by the host where an employee used Lovable to build an 'internet' quickly and affordably.
More from All-In Podcast
View all 432 summaries
52 minThe Trillion-Dollar Industries AI Is Disrupting: Voice, Law & the End of the Billable Hour
103 minMore Trillion Dollar IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts
64 minOpen Source Wins, AGI Is Here, and Scorsese’s AI Toolkit with CEOs of Cerebras & Black Forest Labs
103 minAI Sovereignty Wars, Palantir-Nvidia Deal, SCOTUS Birthright Ruling, Newsom’s CA Budget Lie
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free