←
Intel Capital7,324 posts
Intel Capital avatar

Intel Capital

@intelcapital

A force multiplier for early stage startups - inspiring and investing in the future of compute.

◷ Joined March 2009▧ 587 Photos and videos
Intel Capital avatar
Intel Capital@intelcapital·Jan 20

35 years of Intel Capital investing in compute and the next wave of transformative technology 🚀 Since 1991, we’ve deployed over $20B into 1,800+ startups, supporting founders whose innovations have shaped industries and set global standards. From landmark exits like @AsteraLabs, @jobyaviation, @VMware, and @RedHat to recent investments in @anyscalecompute, @fieldai_, @Figure_robot, @Minio, @runpod, and many more, we continue shaping the future of compute. Our focus remains on #AI, #deeptech, cloud-native infrastructure, advanced manufacturing, and the broader foundry ecosystem. Programs like the Embedded Expert Program (ExP) and global portfolio events provide founders with the expertise, resources, and networks to scale faster and smarter. The journey continues, supporting founders and technologies that will define the future of compute. More on our journey and vision in Anthony Lin's blog: intelcapital.com/intel-capit…

Post attachment
Intel Capital avatar
Intel Capital@intelcapital·Nov 17

CEO of @LightbitsLabs Eran Kirzner joined Code Story podcast's Noah Labhart to discuss how the company’s #NVMe over TCP storage background led it into #AI #inference. They cover breaking the GPU memory wall, raising effective GPU utilization from roughly 30% to as high as 80%, and building products on a six-month MVP cycle. Listen to the full episode below:

Buying more GPUs just for more memory? Eran Kirzner explains how Inferra™ by Lightbits extends GPU memory over the network, lifting utilization from ~30% to 80%. On the Code Story podcast. 🎧 codestory.co/episodes/e13-bo… #AIInference #GPU #KVCache #LLM #AIInfrastructure #GPUMemory

Intel Capital avatar
Intel Capital@intelcapital·Nov 16

New research from @Security_Score's STRIKE team finds internet-exposed Copeland XWEB PRO refrigeration controllers are being hijacked for use in a suspected #spam botnet and as proxy exit points. Three of nine email-sending IPs were confirmed as controllers, with the overlap in their targets suggesting a single coordinated operator. Read more from @helpnetsecurity below:

📰 Help Net Security covered research from SecurityScorecard’s STRIKE Threat Intelligence Team on hijacked refrigeration controllers. Here’s what our team found: 🧊 Copeland XWEB PRO controllers sit inside a suspected spam botnet 🌐 Attackers also use exposed units as proxy exits and tunnel gateways 🔧 Unpatched controllers give attackers a Linux device that no security team watches “Refrigeration control software is internet-connected but often goes without security updates, either because updates require taking equipment offline or because the risk is overlooked,” STRIKE researchers wrote. 👉 Read the Help Net Security coverage and see why exposed controllers go without updates: helpnetsecurity.com/2026/11/… #cybersecurity #threatintel #TPRM #cyberrisk #vendorriskmanagement #thirdpartyrisk #fridgehacking

Intel Capital avatar
Intel Capital@intelcapital·Nov 13

ICYMI, @FieldAI_’s #physicalAI software is now deployed at scale on @KajimaCorp job sites. Its Field Foundation Models act as a general-purpose “brain” that can run on any robot, giving it autonomous navigation without GPS or external connectivity. Since signing an enterprise agreement, FieldAI has expanded to 10+ Kajima sites in Japan and 2,000+ autonomous missions. Read the full release below:

@KAJIMA_CORP announced that FieldAI-powered robots are in full-scale deployment across its construction sites in Japan, with deployments and applications continuing to expand. Field Foundation Models™ serve as a general-purpose robot “brain,” enabling diverse missions across multiple robot types. Kajima has put that intelligence to work in some of the most demanding conditions: into a collapsed roadway, up to a hazardous tunnel face, and even through the air on a crane. The same “brain” can extend to new applications, from large machines for earthwork to humanoids for manipulation. This is what the future of construction looks like. Read the full announcement here: kajima.co.jp/english/news/in…

Post attachment
Intel Capital avatar
Intel Capital@intelcapital·Nov 13

Congratulations to @FigureAI and @higgsfield_ai, which were named to the inaugural AI 100, a ranking of the 100 private companies defining the age of AI from @BessemerVP and @Newcomer. Figure (#55) is building humanoid robots to address labor shortages and unsafe work, and Higgsfield (#58) is building an AI video platform for cinematic content generation. See the full list below: ai100.newcomer.co/

Intel Capital avatar
Intel Capital@intelcapital·Nov 12

Congratulations to @AyarLabs on winning Demo of the Year at the 2026 TSMC North America Technology Symposium, as voted by visitors to the “Innovate with TSMC!” Demo Zone. The demo paired two AI accelerators, each with eight TeraPHY optical engines, in a chassis-level server reference design built with @WiwynnCorp. It delivers 100+ Tbps of bandwidth per accelerator and shows how #opticalinterconnect can extend #AI scale-up beyond the rack. Read more below:

We are proud to be the recipient of TSMC's Demo of the Year Award. Together with TSMC and its ecosystem, we’re enabling a new class of AI infrastructure where optical connectivity extends scale-up beyond the boundaries of the rack. Full press release: bit.ly/4BPjf5r

Intel Capital avatar
Intel Capital@intelcapital·Nov 12

Congratulations to @MetalenzNews who has acquired PxE Imaging. PxE’s Holocoder captures the more than half of incoming light that standard color cameras discard, and adds native depth from the same RGB image with a single camera and no sensor fusion. Metalenz will build on foundry infrastructure that already exists, backed by a supply chain that has shipped 300M+ units. Read more from @WIRED’s Julian Chokkattu below:

WIRED@WIREDNov 11

Metalenz just acquired PxE to ditch the 50-year-old color filter that has held back smartphone night photography. buff.ly/dupik8O

Intel Capital avatar
Intel Capital@intelcapital·Nov 10

Today, @AMFG_ai launched Kurf., a part cost estimation plugin for @Autodesk Inventor. It gives design engineers instant should-cost analysis, DFM warnings and run-time estimates inside the CAD workspace, rather than after a design has gone out to estimators or vendors for quoting. Read more below: amfg.ai/kurf

Intel Capital avatar
Intel Capital@intelcapital·Nov 9

Does #agenticAI need specialized CPUs? @SambaNova VP of Product and Core Tech Anton McGonnell (@aton2006) told @FierceNetwork_ it doesn’t. Agents are asking CPUs to do what computers have always done, like querying databases, compiling code, and running tools, just at far greater volume. Read more below:

Does agentic AI need a new kind of CPU? Our VP of Product and Core Tech, @aton2006, told @FierceNetwork_ it doesn’t. Agents are asking CPUs to do what computers have always done (query databases, compile code, run tools) just a lot more of it. Read the full piece: bit.ly/4iEyci3

Post attachment
Intel Capital avatar
Intel Capital@intelcapital·Nov 9

In an interview with @helpnetsecurity, @immuta CISO Michael Scott predicts major enterprises could face critical #AI incidents within the next 18 months. His concern is less about the models themselves than about organizations giving AI agents access to sensitive #data before they’ve worked out how to control it. His view: agents shouldn’t inherit everything a user can access. Permissions should match the task, expire when the work is done, and stay traceable across every agent and tool involved. Read the full interview below:

Major enterprises could face critical AI incidents within the next 18 months. That’s the prediction from Immuta CISO Michael Scott in an interview with @helpnetsecurity. But his bigger concern is what’s driving that risk: organizations are giving AI agents access to sensitive data before they’ve worked out how to properly control that access. Agents shouldn’t automatically inherit everything a user can access. They need permissions that match the task, expire when the work is done, and remain traceable through every agent and tool involved. Read the full interview: ow.ly/Byfc51055F2

Intel Capital avatar
Intel Capital@intelcapital·Nov 6

Congrats to @twelvelabs Co-founder and CTO Aiden Lee on being named to @MITTechnology Korea’s AI 30, a list of 30 people shaping the future of Korean AI 👏 Aiden helped develop Twelve Labs’ Marengo and Pegasus #VideoAI models. See more about the recognition here: technologyreview.kr/the-ai-3…

Intel Capital avatar
Intel Capital@intelcapital·Nov 6

On @NYSE Wired’s #AIFactories series, @AI21Labs Co-founder and CEO @origoshen joined @theCUBE's Gemma Allen to discuss the next wave of #EnterpriseAI: companies owning specialized models that cut costs and set them apart. Watch the full interview here: youtube.com/watch?v=pwYeC2y7…

Intel Capital avatar
Intel Capital@intelcapital·Nov 5

VP of Core Tech at @SambaNovaAI, Anton McGonnell, sat down with @DW_DCS to discuss the engineering behind the SN50 processor. They cover how chip architecture is evolving for next-generation #inference workloads, why software optimization and silicon design can no longer be treated separately, and how manufacturing scale shapes what’s possible in AI #semiconductor design. Watch the full interview below:

What does it actually take to build a next-generation AI inference chip? @aton2006 sat down with @dw_dcs to go inside the SN50: the architecture decisions, the software optimization behind it, and why manufacturing scale is now part of the design problem. Watch the full discussion here: bit.ly/4z4VIL7

Intel Capital avatar
Intel Capital@intelcapital·Nov 5

Bryan Cantrill (@bcantrill) of @OxideComputer asks why #Rust and code still matter when LLMs can write it. In this Rust NYC talk, he traces Rust’s infrastructure-first design to explain what we should preserve as software development changes, and where LLMs can deepen understanding rather than replace it. Watch the full talk and Q&A below:

For those who have been asking, the @rust_nyc talk with @bcantrill is now live on YouTube! youtu.be/5D7QPOIZ7gY

Intel Capital avatar
Intel Capital@intelcapital·Nov 2

In @CounterPointTR's latest report on Optics, Interconnects and Networks, Eliyan was featured in an analysis from Michael Kanellos. The report looks at Eliyan’s approach to the #interconnect bottleneck in #AI infrastructure, where data transport between chips is increasingly a limit on system performance. Read more below: counterpointresearch.com/en/…

Intel Capital avatar
Intel Capital@intelcapital·Nov 2

Last week, @SambaNovaAI’s Vansanth Mohan (@v_mohan_) joined @EvanKirstel to discuss the economics of #AgenticAI. Request count isn’t what breaks agent budgets. Output tokens per task, multiplied by generation speed, is. Because agents make serial calls, latency and cost compound together, which is why SN50 is built around decode. Watch the full conversation below:

The request count isn’t what breaks agent budgets. It’s output tokens per task, times how fast you can generate them. Serial calls mean latency and cost compound together, which is why SN50 is built around decode. @v_mohan_ walks through the math with @EvanKirstel in their conversation. Check it out! ⬇️