▶ 14:32 · AI Infrastructure · Hyperscalers building custom silicon need merchant interconnect to challenge Nvidia

episode briefing

Inside Kalshi’s $4B+ in Annualized Revenue, Chip Infrastructure Startup Eliyan Hits Unicorn Status

2026-08-13 · 16 company · 11 thematic

sentiment

7 bull2 bear7 neu

speakers

Co-founded Eliyan, a chiplet interconnect startup that raised $145M at $1B valuation; focuses on die-to-die and optical rack-to-rack connectivity for AI systems.

patrick soheili

Co-founded Eliyan; previously worked in semiconductor industry; provides deep technical perspective on memory bottleneck, hyperscaler custom silicon, and interconnect landscape.

yuechi yang

Covers crypto and prediction markets; broke exclusive story on Kalshi's financials and fundraising.

leo schwartz

Covers intersection of AI, crypto, and Washington policy; reported on David Sacks' return to Craft Ventures and White House AI policy dynamics.

laura bratton

Writes the Applied AI newsletter for The Information, reporting on enterprise AI adoption, coding agents, and frontier model competition.

now playing · AI Infrastructure

Prediction Marketsmixedscore 6/10yuechi yang

Kalshi surpasses Polymarket in US market share amid regulatory uncertainty

Kalshi overtook Polymarket as the largest US prediction market by moving faster into regulated US markets, but faces Supreme Court risk on sports contracts and state-level taxation (e.g., N…

Memory & Storagetailwindscore 9/10patrick soheili

Structural memory shortage: three suppliers, concentrated demand, multi-year fab lead times

HBM and advanced memory supply is bottlenecked at Samsung, SK Hynix, and Micron; hyperscaler demand is insatiable and concentrated, while new leading-edge memory fabs require years and bill…

AI Hardware & Chip Architecturetailwindscore 9/10ramin farjadrad

AI shifts from chip problem to system problem: connectivity becomes the critical bottleneck

As Moore's Law slows, AI scaling requires multi-chiplet packages and rack-scale systems; the bottleneck has moved from compute to interconnect (die-to-die, chip-to-chip, rack-to-rack), crea…

Semiconductorstailwindscore 8/10patrick soheili

Hyperscalers vertically integrating custom silicon to escape Nvidia tax and optimize workloads

Every major cloud provider (Google, AWS, Microsoft, Meta, Chinese giants) is investing heavily in proprietary AI accelerators (TPU, Trainium, Maia, MTIA, etc.) to reduce Nvidia dependency a…

AI Hardware & Chip Architecturetailwindscore 8/10ramin farjadrad

Connectivity becomes the bottleneck as AI shifts from chip to system problem

AI infrastructure is transitioning from a chip-level optimization problem to a system-level connectivity problem; chiplet-based architectures require die-to-die, chip-to-chip, and rack-to-r…

Memory & Storagetailwindscore 8/10patrick soheili

Structural memory supply constraint with only three capable suppliers

HBM and advanced memory production is concentrated in Samsung, SK Hynix, and Micron; building new leading-edge fabs takes years and billions, while AI-driven demand is insatiable, creating…

AI Infrastructuretailwindscore 7/10patrick soheili

Hyperscalers building custom silicon need merchant interconnect to challenge Nvidia

Major cloud providers (Google, AWS, Microsoft, Meta, Chinese giants) are developing custom AI accelerators to reduce Nvidia dependency, but require high-performance interconnect IP (like El…

Venture Capitalmixedscore 6/10leo schwartz

VC funds shifting to larger later-stage vehicles as seed rounds inflate

Craft Ventures' move to a $1B sole fund focused on Series A+ (abandoning seed) reflects industry-wide trend where seed rounds now resemble Series C, pushing firms to consolidate funds and d…

Government Technologytailwindscore 7/10laura bratton

US federal agencies accelerating AI adoption: USDA awards $300M to Palantir for full application rebuild

Government IT modernization is moving beyond pilots to large-scale AI-driven application overhauls; Palantir's $300M USDA contract to rebuild farmer acreage reporting demonstrates its emerg…

Enterprise Softwareheadwindscore 7/10laura bratton

Large enterprises beginning to displace incumbent SaaS for specialized AI tools

USDA's shift from Salesforce to C3.ai for a specific AI workload — despite higher cost — marks the first observed large-enterprise pullback from a core SaaS vendor, threatening the pricing…

Government Technologytailwindscore 7/10laura bratton

US federal agencies accelerating AI adoption with Palantir as key beneficiary

USDA's $300M Palantir contract for a complete AI overhaul of its acreage reporting system signals broader government AI modernization, with Palantir positioned as a primary platform for cus…

←

▶ 14:32 · AI Infrastructure · Hyperscalers building custom silicon need merchant interconnect to challenge Nvidia

episode briefing

The Information

Inside Kalshi’s $4B+ in Annualized Revenue, Chip Infrastructure Startup Eliyan Hits Unicorn Status

2026-08-13 · 16 company · 11 thematic

sentiment

7 bull2 bear7 neu

speakers

ramin farjadrad

Co-founded Eliyan, a chiplet interconnect startup that raised $145M at $1B valuation; focuses on die-to-die and optical rack-to-rack connectivity for AI systems.

patrick soheili

Co-founded Eliyan; previously worked in semiconductor industry; provides deep technical perspective on memory bottleneck, hyperscaler custom silicon, and interconnect landscape.

yuechi yang

Covers crypto and prediction markets; broke exclusive story on Kalshi's financials and fundraising.

leo schwartz

Covers intersection of AI, crypto, and Washington policy; reported on David Sacks' return to Craft Ventures and White House AI policy dynamics.

laura bratton

Writes the Applied AI newsletter for The Information, reporting on enterprise AI adoption, coding agents, and frontier model competition.

now playing · AI Infrastructure

Prediction Marketsmixedscore 6/10yuechi yang

Kalshi surpasses Polymarket in US market share amid regulatory uncertainty

Kalshi overtook Polymarket as the largest US prediction market by moving faster into regulated US markets, but faces Supreme Court risk on sports contracts and state-level taxation (e.g., N…

Memory & Storagetailwindscore 9/10patrick soheili

Structural memory shortage: three suppliers, concentrated demand, multi-year fab lead times

HBM and advanced memory supply is bottlenecked at Samsung, SK Hynix, and Micron; hyperscaler demand is insatiable and concentrated, while new leading-edge memory fabs require years and bill…

AI Hardware & Chip Architecturetailwindscore 9/10ramin farjadrad

AI shifts from chip problem to system problem: connectivity becomes the critical bottleneck

As Moore's Law slows, AI scaling requires multi-chiplet packages and rack-scale systems; the bottleneck has moved from compute to interconnect (die-to-die, chip-to-chip, rack-to-rack), crea…

Semiconductorstailwindscore 8/10patrick soheili

Hyperscalers vertically integrating custom silicon to escape Nvidia tax and optimize workloads

Every major cloud provider (Google, AWS, Microsoft, Meta, Chinese giants) is investing heavily in proprietary AI accelerators (TPU, Trainium, Maia, MTIA, etc.) to reduce Nvidia dependency a…

AI Hardware & Chip Architecturetailwindscore 8/10ramin farjadrad

Connectivity becomes the bottleneck as AI shifts from chip to system problem

AI infrastructure is transitioning from a chip-level optimization problem to a system-level connectivity problem; chiplet-based architectures require die-to-die, chip-to-chip, and rack-to-r…

Memory & Storagetailwindscore 8/10patrick soheili

Structural memory supply constraint with only three capable suppliers

HBM and advanced memory production is concentrated in Samsung, SK Hynix, and Micron; building new leading-edge fabs takes years and billions, while AI-driven demand is insatiable, creating…

AI Infrastructuretailwindscore 7/10patrick soheili

Hyperscalers building custom silicon need merchant interconnect to challenge Nvidia

Major cloud providers (Google, AWS, Microsoft, Meta, Chinese giants) are developing custom AI accelerators to reduce Nvidia dependency, but require high-performance interconnect IP (like El…

Venture Capitalmixedscore 6/10leo schwartz

VC funds shifting to larger later-stage vehicles as seed rounds inflate

Craft Ventures' move to a $1B sole fund focused on Series A+ (abandoning seed) reflects industry-wide trend where seed rounds now resemble Series C, pushing firms to consolidate funds and d…

Government Technologytailwindscore 7/10laura bratton

US federal agencies accelerating AI adoption: USDA awards $300M to Palantir for full application rebuild

Government IT modernization is moving beyond pilots to large-scale AI-driven application overhauls; Palantir's $300M USDA contract to rebuild farmer acreage reporting demonstrates its emerg…

Enterprise Softwareheadwindscore 7/10laura bratton

Large enterprises beginning to displace incumbent SaaS for specialized AI tools

USDA's shift from Salesforce to C3.ai for a specific AI workload — despite higher cost — marks the first observed large-enterprise pullback from a core SaaS vendor, threatening the pricing…

Government Technologytailwindscore 7/10laura bratton

US federal agencies accelerating AI adoption with Palantir as key beneficiary

USDA's $300M Palantir contract for a complete AI overhaul of its acreage reporting system signals broader government AI modernization, with Palantir positioned as a primary platform for cus…