La CEO de AMD, Lisa Su, acaba de acabar con la caja de IA de $4,000 de Nvidia con una lonchera de $1,499. Subió al escenario, la sostuvo en una mano y ejecutó en vivo un modelo de 235 mil millones de parámetros. Sin centro de datos. Sin nube. Sin GPU alquilada. El chip en su interior es algo que nadie vio venir. El Ryzen AI Max 395 de AMD es el primer silicio x86 donde la CPU y la GPU comparten los mismos 128 GB de memoria. Ese solo truco permite que un escritorio ejecute modelos que antes necesitaban un rack de servidores. De esos 128 GB, Linux le da al GPU 110 GB para jugar. Para contextualizar, una RTX 5090 te da 32 GB. Una 4090 te da 24. Esta caja te da más del triple que cualquiera de ellas, en un chasis del tamaño de un libro de bolsillo grueso. El benchmark que rompió la sala: este chip superó a una Nvidia RTX 5080 por más de 3x en inferencia de DeepSeek R1. Una lonchera de $1,499 superando a una tarjeta gráfica discreta de $1,000 en una carga de trabajo real de IA. Nvidia pasó una década convenciendo al mundo de que necesitabas su hardware para IA seria. AMD acaba de poner eso en un escritorio por la mitad del precio. Aquí está lo que nadie te está diciendo. Un usuario intensivo de IA ahora paga $200 por Claude Code Max, $200 por ChatGPT Pro, $20 por Cursor, $20 por Gemini. Eso son $5,280 al año saliendo de tu cuenta. La caja se paga sola en 9 meses y luego corre gratis por el resto de su vida. Instala Ollama. Descarga Qwen3 235B. Apunta Claude Code a localhost. La misma interfaz que ya usas, excepto que ahora nada sale de tu máquina, nada cuesta por solicitud y ninguna empresa limita tu uso a las 3 de la mañana cuando por fin tienes tiempo para construir. Este es el momento en que todas las suscripciones de IA se vuelven opcionales. Los abogados dejan de temer fugas de OpenAI. Los desarrolladores dejan de mirar el medidor de tokens. Los fundadores dejan de alquilar H100s para prototipos que nunca se envían porque la factura los asustó. Las primeras mil personas en descifrar esto poseerán los próximos dos años de consultoría de IA privada.

$IONQ Every Scout starts with two words — BE PREPARED — and most never need them to mean more than a packed bag. The few who build a life around them, who treat readiness as a discipline rather than a slogan and carry it into service where the stakes are real, are the ones from whom Tier-1 operators are eventually drawn: not the loudest, not the best-equipped, but the most reliably ready when it counts — and that is exactly the distance my new report, The Delivery Verdict: IonQ's Tier-1 Mentality, measures in IonQ. Here's the thesis, and I'm not going to soften it: as global enterprise crosses from quantum experimentation into quantum commercialization, the entire game stops being about who has the flashiest roadmap and becomes about who actually delivers — on time, ahead of schedule, when sovereign capability and real money are on the line — and on that measure IonQ isn't competing for the lead, it is the lead. THE BIG QUESTION IS THIS: Does @IonQ_Inc fulfill its stated commitments and are they on-time, ahead of schedule, or behind, and what does this mean for their business operations and roadmap? In essence, can we TRUST IonQ to deliver what we many expect will be delivered in the future? Most of us know, milestones hit early. Guidance beaten five quarters running. An integrated national-security full stack — computing, networking, sensing, security — delivered at a breadth no competitor anywhere is close to matching at the moment, while the rest of the field is still talking. We get it. You get it. But can the momentum sustain itself? This is the moment that separates the operators from the aspirants, and the report documents — evidence-tagged, primary-sourced, open conditions intact — why IonQ has already crossed that line. IonQ has developed in many ways characteristics of a Tier-1 Operator. Bottom line: this is the report that sets everything apart and tells you why. quantumtechintegration.blogs…

#OCUGEN 💎⛏️ 💎⛏️ #LETSGOOO

IONQ said this to a lot of people and they were lying

$IONQ IonQ posted a paper titled "Breakeven demonstration of quantum low-density parity-check codes." Eleven IonQ authors, including Edwin Tham, Michael Goldman, Kenneth Wright, and Nicolas Delfosse. Quantum computers work with individual atoms or other tiny physical systems as their basic units. Each unit is a physical qubit, and physical qubits are extremely fragile. Small disturbances cause errors that accumulate fast. The way around this is quantum error correction. Many physical qubits group together to encode one logical qubit, a virtual unit more reliable than any of the physical qubits underneath. The logical qubit is what actually does the computation. The physical qubits constantly check each other for errors and correct them. A code is the specific recipe for how those physical qubits work together. Breakeven is the milestone where the logical qubit lives longer than the physical qubits inside it. The error correction is fixing errors faster than the system is adding new ones from the overhead of running the code. Before breakeven, error correction is theoretical. After breakeven, it works. Google reached breakeven last year on superconducting hardware. Superconducting qubits are tiny circuits cooled to near absolute zero. IBM, Google, and several others build on this approach. Today IonQ reached breakeven on trapped-ion hardware. Trapped ions are individual atoms held in place by electric fields. No deep cooling needed. IonQ ran nine different error-correcting codes across three different code families. All on the same 40-ion device. No hardware changes between codes. The best logical qubit lived 3.95 seconds. The physical qubits underneath it lived 3.3 seconds. The logical qubit outlived the physical qubits it was built from. The technical innovation is called OMG. The system reads out individual qubits in the middle of a computation, then resets them, without physically moving any ions around. Other trapped-ion systems spend most of their runtime moving ions back and forth between different zones of the chip. They also dedicate up to half their qubits just to cooling the ones doing computation. IonQ's approach eliminates both. The paper also makes a direct comparison to a recent IBM result. Same code and same qubit counts. IonQ's logical error rates came in 4 to 9 times lower. IBM had built a chip with custom structures just to run that one code. IonQ ran nine codes on the same standard device. IonQ has joined the small group of companies that have demonstrated fault-tolerant error correction at breakeven. Google did it last year on superconducting hardware. IonQ did it today on trapped ions. Where the comparison is direct, IonQ's hardware comes out ahead. The error rates were 4 to 9 times lower than IBM's on the same code. IonQ ran nine codes on its standard hardware where IBM required a custom chip just for one. And other trapped-ion systems carry transport and cooling overhead that IonQ's design avoids. The architectural bet is paying off in the field's most demanding experimental test.