Investment Research
AI Stock Universe: Top 5 Candidates for Jan 1, 2029 Return
A funnel from the 513-stock AI universe to a final deployable top five. This assumes AGI by 2027 and a 2029 world where compute, power, networking, packaging, and physical deployment bottlenecks matter more than generic "AI feature" exposure.
Important correction: the right question is not "which AI companies are best?" It is "which stocks have the best return from today's price?" Those are different. NVIDIA, TSMC, Broadcom, Microsoft, Google, and Meta may be among the biggest AI winners and still fail to be top percentage-return stocks from current market caps.
Final Answer
1. SEI - Solaris Energy Infrastructure - $7.2B
Best valuation-adjusted bridge-power candidate. Mobile gas turbine capacity can monetize while utilities, transmission, and nuclear are too slow for 2026-2029 demand.
2. LMB - Limbach Holdings - $0.9B
Cheapest direct physical-infrastructure candidate in the pool. It is small and lower capacity, but valuation is sane enough that the upside does not require perfection.
3. IESC - IES Holdings - $13.4B
Direct data-center electrical exposure with better valuation support than POWL/CRDO. It is not cheap, but it is underwritable.
4. APLD - Applied Digital - $12.6B
Still one of the highest-upside powered-campus bets, but it is explicitly expensive and execution-heavy. This is an option on powered AI campuses, not a cheap stock.
5. CAMT - Camtek - $9.4B
Best advanced-packaging bottleneck pick. It is expensive, but the cap is still small enough that HBM/CoWoS inspection intensity can move the stock.
How I Am Interpreting "Highest Returns"
For a "bunch of capital," I would rank by probability-weighted, deployable return, not by the biggest theoretical 20x. That means I penalize pre-revenue and tiny-cap names even when their upside is enormous.
Valuation-adjusted deployable list: SEI, LMB, IESC, APLD, CAMT.
Raw moonshot list: BTBT, POET, ASYS, SERV, QUBT. These can outperform the final five, but I would not make them core capital-deployment positions without much deeper single-name diligence.
The valuation layer demotes POWL and CRDO from the final five. They stay in the 50-stock winner pool because they can still be the right answer if switchgear or interconnect scarcity lasts longer than the market expects.
Valuation Reality Check
This is the missing layer from the first version. I pulled a live yfinance snapshot on 2026-05-08 and used it as a sanity check against the universe market caps. These numbers move, and they are not a substitute for full single-name models, but they are enough to catch the obvious problem: several good AI names are priced for near-perfect execution.
| Ticker | Mkt Cap | EV/Sales | EV/EBITDA | Fwd P/E | Rev Growth | Valuation Read |
|---|---|---|---|---|---|---|
| LMB | $0.9B | 1.4x | 14.6x | 14.1x | 4% | Actually cheap relative to the rest of the AI physical-infra universe. Lower scale and lower direct AI purity, but valuation gives room for mistakes. |
| IESC | $13.3B | 3.6x | 27.4x | n/a | 17% | Not cheap, but underwritable if data-center electrical and Infrastructure Solutions margins persist. Better valuation support than POWL. |
| SEI | $6.8B | 8.5x | 23.1x | 26.8x | 55% | Expensive but not absurd for a high-growth bridge-power scarcity asset. Needs contracted MW and EBITDA conversion. |
| APLD | $11.8B | 43.4x | 998x | neg. | 139% | Very expensive on current revenue. This is a project-development option on powered campuses; it only works if signed leases and financing scale fast. |
| CAMT | $9.6B | 19.0x | 65.6x | 47.1x | 9% | Expensive for a cyclical semicap name. Still survives because the market cap is small and the packaging bottleneck can overwhelm historical cyclicality. |
| POWL | $11.3B | 9.5x | 46.3x | 44.7x | 7% | Good business, bad entry if margins normalize. This valuation requires backlog and margins to stay structurally elevated. |
| CRDO | $34.8B | 31.3x | 95.6x | 34.2x | 202% | Pricey, but revenue growth is real. It can still win, but valuation makes it less attractive than the physical-infra names. |
| ALAB | $34.2B | 33.1x | 141.8x | 47.8x | 93% | Excellent business, priced for excellence. Keep in pool, but do not pretend valuation is forgiving. |
| POET | $1.7B | 1058x | neg. | n/a | off tiny base | Not a valuation. This is a venture option on silicon photonics commercialization. |
| BTBT | $0.6B | 6.7x | neg. | n/a | 25% | Small enough to matter, but current financials are weak. Only belongs in the winner pool because the AI hosting pivot can re-rate the whole company. |
Funnel: 513 -> 250 -> 125 -> 60 -> 30 -> 10 -> 5
The main decision rule: AGI should increase the company's economic value by 2029, and the starting price should leave room for a large percentage return. I am intentionally harsh on companies where AI is real but already obvious.
| Level | Count | Question | Kept | Cut |
|---|---|---|---|---|
| Universe | 513 | All names listed in the AI Stock Universe page. | Hyperscalers, semis, networking, infra, power, software. | None yet, except treating GOOG/GOOGL as duplicate economic exposure. |
| Direct AI sensitivity | ~250 | Does AI materially change revenue, margins, asset value, or strategic scarcity by 2029? | Compute supply chain, data-center infrastructure, power, interconnect, advanced packaging, a few AI-native apps. | Payment networks, broad pharma, generic banks, media, most industrials where AI is only an internal efficiency tool. |
| Positive AGI exposure | ~125 | Is AGI more likely to help than cannibalize the business model? | Physical bottlenecks, proprietary manufacturing, mission-critical infrastructure, scarce powered sites. | IT services, seat-based SaaS with weak moats, ad verification, labor-arbitrage businesses, generic "AI feature" apps. |
| Return from today's cap | ~60 | Can this plausibly 2x-5x by Jan 2029 without requiring absurd market cap math? | Sub-$50B names with real bottleneck exposure; a few larger names with exceptional quality. | Most mega-cap winners: NVDA, TSM, AVGO, MSFT, GOOGL, AMZN, META, ASML, AMAT, LRCX. |
| Bottleneck durability | ~30 | Is the constraint hard to solve quickly, even with smarter AI? | Power, switchgear, data-center construction, HBM/CoWoS inspection, SerDes/cables, defense autonomy. | Most software wrappers, small quantum names, autonomy names with weak commercial proof, broad commodity exposure. |
| Underwriting shortlist | 10 | Does the thesis have both upside and a visible path to validation before 2029? | SEI, LMB, IESC, APLD, CAMT, CRDO, POWL, ALAB, POET, LEU. | BTBT/ASYS/SERV/QUBT as too speculative for core sizing; MOD/TLN/AGX as good but less asymmetric; DUOL/NICE/S as software near misses. |
| Final deployable top 5 | 5 | Where would I most want incremental capital if the mandate is 2029 AI-return maximization? | SEI, LMB, IESC, APLD, CAMT. | POWL and CRDO because valuation is harsher than the first pass reflected; ALAB due to perfection pricing; POET due to venture-style execution risk; LEU due to timing risk. |
Final Top 5: Scenario Math and Kill Criteria
These are rough market-cap outcome ranges, not precise valuations. The point is to make the return logic explicit: what has to be true, what breaks the thesis, and why the stock can outrun today's price.
| Rank | Ticker | Category | Current Cap | Why It Can Win | Bear / Base / Bull Cap by 2029 | Kill Criteria | Score |
|---|---|---|---|---|---|---|---|
| 1 | SEI Solaris Energy Infrastructure |
Power | $7.2B | Bridge power is the most urgent 2026-2029 constraint, and SEI is small enough for contracted MW growth to matter. The valuation is expensive but still tied to real EBITDA, not pure concept value. | Bear: $3B-$4B if contracts are transient or leverage bites. Base: $18B-$25B if multi-year data-center power contracts convert. Bull: $40B-$60B if mobile power becomes a standard hyperscaler procurement path. |
Data-center bookings stay anecdotal; contracts are short-duration or low-margin; emissions/permitting blocks deployments; debt issuance outruns contracted cash flow. | 8.8 |
| 2 | LMB Limbach Holdings |
AI Infra | $0.9B | Small mission-critical mechanical/electrical contractor. The AI linkage is less pure than APLD/SEI, but the valuation is far more forgiving, which matters for actual returns. | Bear: $0.4B-$0.6B if growth stays ordinary and margins compress. Base: $2B-$3B if mission-critical backlog grows and the multiple expands modestly. Bull: $5B-$8B if it becomes a recognized small-cap data-center MEP compounder. |
Backlog growth is not tied to mission-critical/data-center work; scale limits large customer wins; margins revert before revenue mix improves. | 8.6 |
| 3 | IESC IES Holdings |
AI Infra | $13.4B | Direct electrical contractor exposure to data-center buildouts plus a higher-margin infrastructure-solutions business. The valuation is elevated but supported by real revenue and profits. | Bear: $7B-$9B if contractor multiples compress and margins normalize. Base: $25B-$35B if data-center electrical demand persists. Bull: $50B-$70B if hidden infrastructure-solutions earnings get valued like a critical electrical-equipment platform. |
Infrastructure Solutions margins normalize; backlog visibility deteriorates; data-center customer concentration causes a revenue air pocket. | 8.3 |
| 4 | APLD Applied Digital |
AI Infra | $12.6B | High-torque AI/HPC data-center developer. Valuation is harsh, but powered campuses can still create a large equity outcome if signed leases and project finance scale. | Bear: $4B-$6B if financing slips or dilution is punitive. Base: $25B-$35B if key campuses lease and finance cleanly. Bull: $60B-$90B if APLD becomes a preferred powered-campus platform. |
Signed customer commitments lag buildout; Macquarie-style financing gets expensive; major construction delays; equity dilution funds survival rather than growth. | 8.0 |
| 5 | CAMT Camtek |
Semiconductors | $9.4B | HBM, CoWoS, chiplets, and advanced packaging make inspection and metrology more valuable. Expensive, but the cap is low enough for packaging scarcity to matter. | Bear: $4B-$6B if packaging capex rolls over. Base: $18B-$25B if inspection intensity keeps rising. Bull: $35B-$50B if advanced packaging stays the bottleneck into 2028. |
HBM/CoWoS capex pauses; customers dual-source aggressively; margins normalize like a cyclical tool vendor; China/geopolitical exposure becomes material. | 7.8 |
Top 10 Shortlist Before Final Cut
This is the set I would keep in active diligence. The difference between ranks 5-10 is not huge; it depends heavily on how much failure probability you are willing to tolerate.
| Rank | Ticker | Cap | Thesis | Why Not Higher |
|---|---|---|---|---|
| 1 | SEI | $7.2B | Mobile turbine bridge power for data-center load before grid interconnect catches up. | Needs durable data-center contracts, not just energy-cycle enthusiasm. |
| 2 | LMB | $0.9B | Mission-critical MEP contractor at a valuation that still leaves room for surprise upside. | AI/data-center purity is weaker than larger infra names; scale can limit opportunity size. |
| 3 | IESC | $13.4B | Electrical contractor with direct data-center buildout exposure and unusually high margins. | Already rerated; contractor multiples can compress quickly if backlog slows. |
| 4 | APLD | $12.6B | Powered AI/HPC campus developer with strong upside if power scarcity persists. | Valuation is extreme; this needs lease/finance execution, not just narrative. |
| 5 | CAMT | $9.4B | Advanced packaging inspection directly tied to HBM/CoWoS yield bottlenecks. | Semicap cycle and valuation can still overwhelm the secular thesis. |
| 6 | CRDO | $36.6B | Active electrical cables and SerDes for AI cluster scale-out. | Excellent growth, but already a large cap for a component supplier. |
| 7 | POWL | $11.7B | Switchgear scarcity across data centers, grid, and LNG. | Current valuation requires sustained high margins; less forgiving than LMB/IESC. |
| 8 | ALAB | $36.7B | PCIe/CXL retimers and smart cable modules for AI servers and composable memory. | Excellent business, but valuation assumes a lot; Broadcom/Marvell competition matters. |
| 9 | POET | $1.5B | Silicon photonics/optical engines. Huge upside if optical I/O becomes the binding cluster bottleneck. | Pre-revenue-style risk: commercialization, dilution, customer validation, manufacturing yield. |
| 10 | LEU | $4.6B | Only US-owned uranium enricher; HALEU scarcity if AI load forces nuclear acceleration. | Policy and nuclear build timelines may not pay off by Jan 2029. |
50-Stock Winner Pool
This is not a buy list. This is the "do not miss the 2029 outlier" pool. The final top five is valuation-adjusted and deployable; this pool is broader and intentionally includes a few speculative names because the actual highest-return stock may be a small, ugly, hard-to-size company that crosses a commercial validation threshold.
If the Jan 1, 2029 winner is a fundamentally plausible AI beneficiary from today's public universe, I expect it to be in this 50. If the winner is a pure fraud, reverse-merger pump, or unrelated meme squeeze, this pool may miss it by design.
| # | Ticker | Bucket | Approx Cap | Valuation Reality | Why It Is In The Pool |
|---|---|---|---|---|---|
| 1 | APLD | Powered Campus | $12.6B | Very expensive; ~43x EV/sales. | One of the cleanest powered AI/HPC campus options. |
| 2 | SEI | Bridge Power | $7.2B | Expensive but real EBITDA; ~23x EV/EBITDA. | Fast mobile power could be a 2026-2029 bottleneck winner. |
| 3 | POWL | Switchgear | $11.7B | Expensive; ~46x EV/EBITDA. | Switchgear scarcity may last longer than the market expects. |
| 4 | IESC | Electrical | $13.4B | Underwritable; ~3.6x EV/sales. | Direct data-center electrical exposure with real profits. |
| 5 | LMB | MEP Contractor | $0.9B | Cheapest quality infra name; ~1.4x EV/sales. | Small enough to be a major winner if mission-critical mix rises. |
| 6 | BTBT | AI Hosting Pivot | $0.6B | Weak current financials; negative EBITDA. | WhiteFiber AI/HPC pivot can re-rate the whole company if real. |
| 7 | BITF | BTC-to-HPC | $1.2B | Speculative; limited current valuation support. | Cheap optionality on mining sites becoming HPC campuses. |
| 8 | CORZ | HPC Host | $7.8B | Expensive on sales; contract-heavy thesis. | CoreWeave host and Leopold activist stake keep it in play. |
| 9 | CIFR | AI Site Option | $9.0B | High multiple; needs Barber Lake execution. | Contracted HPC site exposure can produce nonlinear equity value. |
| 10 | WULF | AI Hosting | $12.7B | Very expensive on current revenue. | Lake Mariner/Core42-style hosting optionality. |
| 11 | IREN | AI Cloud Pivot | $20.2B | Expensive; ~26x EV/sales. | Owns power/site assets and is moving toward AI cloud economics. |
| 12 | NBIS | GPU Cloud | $49.5B | Very expensive; growth/capacity story. | Nebius can be a major non-US AI cloud winner if contracted capacity scales. |
| 13 | CRWV | GPU Cloud | $72.9B | Leveraged; ~18x EV/sales. | Largest pure-play GPU cloud; could still compound if backlog converts. |
| 14 | MOD | Cooling | $14.5B | Not cheap; ~36x EV/EBITDA. | Airedale gives direct data-center cooling exposure. |
| 15 | AAON | Cooling | $8.0B | Expensive; ~44x EV/EBITDA. | Custom HVAC can win if liquid/high-density cooling demand spreads. |
| 16 | SPXC | Cooling | $10.7B | More reasonable; ~22x EV/EBITDA. | Cooling towers and thermal infrastructure exposure. |
| 17 | AGX | Gas EPC | $10.2B | Expensive; project-cycle risk. | Gas-fired EPC can win if AI needs fast dispatchable power. |
| 18 | BE | Fuel Cells | $81.2B | Very expensive; ~30x EV/sales. | Onsite power can be valuable if grid queues remain brutal. |
| 19 | TLN | Nuclear Site | $18.7B | Valuation more grounded; low forward P/E. | Scarce nuclear co-location asset near AI load. |
| 20 | LEU | HALEU | $4.6B | Expensive but strategic. | Only US-owned uranium enricher; nuclear acceleration option. |
| 21 | OKLO | SMR | $13.8B | Concept valuation; no commercial revenue. | Could be a narrative winner if AI nuclear procurement accelerates. |
| 22 | SMR | SMR | $4.4B | Very expensive relative to revenue. | NRC-approved SMR design creates event-driven upside. |
| 23 | NNE | Microreactor | $1.5B | Lottery ticket. | Small-cap microreactor/HALEU transport narrative coverage. |
| 24 | LTBR | Nuclear Fuel | $0.5B | Lottery ticket. | Advanced nuclear fuel option with small-cap torque. |
| 25 | CAMT | Packaging | $9.4B | Expensive; ~19x EV/sales. | Advanced packaging inspection can become a key AI compute bottleneck. |
| 26 | CRDO | Interconnect | $36.6B | Expensive but fast growth. | Active electrical cables/SerDes for AI cluster scale-out. |
| 27 | ALAB | Interconnect | $36.7B | Excellent but priced for perfection. | PCIe/CXL retimers and smart cable modules for AI racks. |
| 28 | POET | Silicon Photonics | $1.5B | Venture option; current sales tiny. | Could be a monster if optical engines win AI sockets. |
| 29 | MTSI | RF/Optical | $23.2B | Expensive; ~24x EV/sales. | High-speed analog and optical components for AI networking. |
| 30 | FORM | Probe Cards | $11.6B | Expensive; ~13x EV/sales. | HBM/accelerator testing complexity should rise. |
| 31 | RMBS | Memory IP | $14.1B | Quality IP economics; not cheap. | Bandwidth scarcity benefits memory interfaces, DDR5/HBM/CXL. |
| 32 | KLIC | Packaging Tools | $4.9B | More reasonable; ~6x EV/sales. | Thermo-compression and advanced assembly optionality. |
| 33 | VECO | WFE Tools | $3.8B | Reasonable for the group; ~5x EV/sales. | Small-cap exposure to advanced packaging/HBM tools. |
| 34 | ONTO | Metrology | $14.6B | Expensive but profitable. | Inspection/metrology including advanced packaging for AI HBM. |
| 35 | NVMI | Metrology | $17.0B | Expensive but high margins. | Optical/X-ray metrology for advanced nodes and HBM. |
| 36 | AMKR | OSAT | $19.1B | Cheaper; ~2.5x EV/sales. | OSAT exposure to advanced packaging buildout. |
| 37 | ASYS | Packaging Microcap | $0.3B | Microcap; mixed financials. | Solder reflow furnaces for chip packaging with tiny-cap torque. |
| 38 | NVTS | Power Semis | $3.9B | Very expensive; weak current revenue. | GaN/SiC power conversion for AI server power supplies. |
| 39 | COHR | Optics | $64.6B | Not cheap; ~10x EV/sales. | Lasers/transceivers for 800G/1.6T AI networks. |
| 40 | KTOS | Defense Autonomy | $11.5B | Expensive; defense option. | Autonomous drones and attritable aircraft are direct AI warfare exposure. |
| 41 | MBLY | Autonomy | $7.6B | De-rated; ~3x EV/sales. | If AGI improves autonomy, Mobileye can re-rate from a depressed base. |
| 42 | OUST | Lidar/Robotics | $1.9B | Speculative but not absurd for growth. | Robotics/autonomy lidar optionality with public-market survival. |
| 43 | AMBA | Edge Vision | $3.3B | Speculative; negative margins. | Edge AI vision SoCs for autos, robotics, and security cameras. |
| 44 | SERV | Robotics | $0.7B | Lottery valuation; ~173x sales. | NVIDIA-linked sidewalk robot option; included only for outlier capture. |
| 45 | TSLA | Robotics/Autonomy | $1.50T | Mega-cap, extremely expensive. | Still in pool because Optimus/FSD/robotaxi could dominate the 2029 narrative. |
| 46 | PL | AI Geospatial | $14.1B | Expensive; negative margins. | Proprietary earth-observation data for defense and AI analytics. |
| 47 | IONQ | Quantum | $19.6B | Very expensive; commercial timing uncertain. | Best-known public quantum option. |
| 48 | QUBT | Quantum Lottery | $2.4B | Absurd valuation; ~1500x sales. | Included only because it can be a 2029 narrative outlier despite low quality. |
| 49 | DUOL | AI Tutor | $4.9B | Reasonable for growth; ~3.8x sales. | AI tutor/content creation is accretive rather than purely cannibalistic. |
| 50 | S | Cyber AI | $5.2B | Moderate sales multiple; still loss-making. | AI increases attack volume and need for autonomous SOC tooling. |
Sector-Level Reasoning
Hyperscalers
Best businesses: GOOGL, MSFT, AMZN, META. Best expected return in the slice: BABA/TCEHY/BIDU/META depending on China risk tolerance.
Why mostly cut: the US leaders are already $1.5T-$5T. They can own the future and still produce lower percentage returns than smaller bottleneck suppliers. China names have more upside but add VIE/geopolitical risk.
Semiconductors and Networking
The obvious winners are real: NVDA, TSM, AVGO, ASML, SK Hynix, MU. The return problem is price. The better 2029 percentage-return setup is in second-order bottlenecks: CAMT, CRDO, ALAB, MTSI, FORM, RMBS, KLIC, POET.
Power and Physical Infrastructure
This is the strongest area. AGI can design better systems, but it cannot instantly build substations, turbines, switchgear, pipelines, data centers, and grid interconnects. SEI, LMB, IESC, APLD, POWL, MOD, TLN, AGX, and LEU are the core watchlist.
Software and Apps
Most software is less attractive than it looks because AI can cannibalize seats and compress workflows. The exceptions need distribution, proprietary data, or clear labor-replacement ROI: DUOL, NICE, S, HUBS, GTLB, ESTC, RDDT, ZS. Only DUOL and S make the 50-pool because the physical bottleneck names have cleaner 2029 return asymmetry.
Important Rejects
| Group | Examples | Why I Cut Them |
|---|---|---|
| Obvious mega-cap AI winners | NVDA, TSM, AVGO, MSFT, GOOGL, AMZN, META, ASML | Great companies, but too much of the 2029 story is capitalized. The market cap math is brutal for top percentage returns. |
| Memory cycle leaders | MU, SK Hynix, Samsung | HBM demand is real, but memory remains cyclical. Current prices look closer to peak-cycle enthusiasm than asymmetric entry. |
| Regulated utilities | AEP, D, DUK, PNW, OGE, SO | AI load growth is real, but regulated returns cap upside. Useful compounders, usually not top-return candidates. |
| Commodity giants | XOM, CVX, COP, FCX, SCCO, BHP, RIO | AI demand helps gas/copper, but exposure is diluted by huge commodity businesses and commodity price cycles. |
| Data-center REITs | EQIX, DLR | Good assets, but REIT mechanics, capital intensity, and current valuations blunt upside. |
| AI software wrappers | AI, SOUN, BBAI, PATH | The label is obvious, but durability is weak. Foundation models and larger platforms can absorb much of the product surface. |
| Quantum lottery tickets | QUBT, RGTI, QBTS, IONQ | Most are too speculative for core capital. IONQ and QUBT stay in the 50-pool only for outlier capture; they are not valuation-supported recommendations. |
What Would Change My Mind
The highest-signal work from here is not more generic AI narrative. It is tracking hard validation points that prove or disprove the bottleneck thesis.
| Ticker | Positive Validation | Negative Validation |
|---|---|---|
| SEI | More multi-year data-center power contracts, higher MW under contract, EBITDA scaling without leverage blowout. | Power bookings remain one-off, emissions/permitting blocks growth, debt raised faster than contracted cash flow. |
| LMB | Mission-critical/data-center backlog becomes explicit, margins hold, and the market starts valuing it as AI physical infrastructure rather than a generic contractor. | Revenue growth stays ordinary, backlog is not data-center linked, or margins revert before the mix shift is proven. |
| IESC | Infrastructure Solutions keeps high margins, data-center electrical backlog grows, and revenue visibility improves. | Infrastructure Solutions margin normalizes, backlog stalls, or customer concentration produces a revenue air pocket. |
| APLD | Signed hyperscaler/neocloud leases, non-punitive project finance, on-time campus energization. | Repeated delays, equity issuance to cover cost overruns, customer concentration without better contract terms. |
| CAMT | HBM/CoWoS order strength, packaging-inspection mix expansion, sustained gross margin. | Packaging capex pause, order cancellations, inspection intensity lower than expected. |
Portfolio Implication
If actually deploying meaningful capital, I would not equal-weight the raw top-five upside names. I would build a core around the deployable bottleneck names and keep the venture-style names as a separate sleeve.
A sane research portfolio for this thesis would look like: 60% valuation-adjusted bottlenecks (SEI, LMB, IESC, APLD, CAMT), 25% expensive-but-real bottlenecks (POWL, CRDO, ALAB, MOD, TLN, LEU), and 15% moonshots (POET, BTBT, ASYS, SERV, QUBT, OUST). That is a research framing, not a final allocation.
Live Charts: Final Five
Charts use the shared site chart component. They require the deployed site or local API proxy to fetch live price history.