The current AI cycle is no longer a narrow contest among foundation-model laboratories. Capital is moving across the entire productive system: model developers, inference and orchestration software, data and compliance applications, robotics, semiconductors, cloud infrastructure, cybersecurity, energy, and data centers. Between April 25 and August 2, 2026, the central question for Alphabet is therefore not simply whether Gemini can win a model race. It is whether Alphabet can command the infrastructure, distribution, developer, and enterprise layers that determine whether AI spending becomes durable revenue.
The most consequential signal is the scale and breadth of private financing. Mistral AI has raised billions in cumulative funding 1,19, including $830 million of debt financing 1,67. Reports with relatively strong corroboration describe a potential €3 billion Series D at a €20 billion valuation 3,5,39,42,67,78. Railway has raised $100 million 2,4,36, LangChain more than $150 million 19, and Freehand $75 million 13,14,93. These transactions demonstrate that investors continue to finance both the mills that produce models and the railways that distribute them.
For Alphabet, this is strategically important. The company is simultaneously a hyperscale infrastructure provider, a model developer through Google DeepMind, a cloud platform, and a major distributor of AI applications. The investment conclusion is constructive but not indiscriminately bullish. Funding supports secular demand for compute and cloud services, potentially benefiting Google Cloud, Alphabet’s TPU ecosystem, data platforms, and enterprise distribution. At the same time, the cycle is becoming more capital-intensive and more dependent on scarce chips, power, financing, and customer conversion. The decisive test is not how much capital enters the system, but how much of it becomes productive, recurring workload.
The Capital Cycle Is Concentrating Around the Full AI Stack
Foundation models remain capital-intensive strategic assets
Mistral’s proposed €3 billion Series D, reportedly at a €20 billion valuation, is supported by multiple reports 3,5,7,39,42,67,78. The financing would follow a €1.7 billion Series C 67 completed less than a year earlier 67, bringing reported cumulative debt and equity financing above $3.5 billion 67. Founded in 2023 as a European alternative to US-dominated foundation-model providers 67, Mistral differentiates itself through European regulatory alignment, open-weight releases, and enterprise partnerships rather than raw scale alone 67.
Mistral develops open-weight models 19,67,78, including Mistral Medium 3.5 for general application development and OCR 4 for document processing and agentic workflows 78. Its target markets include manufacturing, financial services, healthcare, defense, document automation, agentic workflows, and applications requiring local deployment or data residency 78. This is a significant strategic distinction. The company is not merely selling model intelligence; it is selling control over where and how that intelligence operates.
The financing itself remains unresolved. Several claims describe the round as reported, potential, or unconfirmed 67,78, and Mistral’s chief executive declined to comment 78. Neither Mistral nor Samsung formally confirmed the reported discussions 67. Samsung’s potential participation of up to €1 billion remains unfinalized 39,42,67. The transaction should therefore be treated as a measure of investor appetite and strategic positioning, not as completed capital. Its reported valuation also creates valuation risk 39, while cross-border execution and closing uncertainty remain explicit concerns 39.
Mistral’s capital requirements reflect the economics of the industry 78. The company depends on US-linked chip suppliers, particularly Nvidia and AMD 78, and faces the operational challenge of deploying models consistently across Azure, Azure Local, and its own facilities, complying with regional laws, securing GPU capacity, and converting benchmark performance into sustainable commercial growth 78. Customers in regulated industries require privacy, data residency, security, and compliance 78. These requirements strengthen the strategic position of cloud platforms capable of supplying compute, deployment controls, regional infrastructure, and enterprise support.
For Alphabet, the lesson is direct: model quality alone will not secure market share. The company must continue to build the infrastructure that allows customers to operate models reliably, economically, and within jurisdictional boundaries. In this new industrial order, the foundation model is the productive asset, but the cloud, network, power supply, and compliance machinery determine whether that asset earns a return.
Application, orchestration, and governance are attracting serious capital
Capital is also moving downstream. Railway raised $100 million 2,4,36, LangChain secured more than $150 million 19, and Stripe’s proposed acquisition of OpenRouter for approximately $10 billion was characterized as an investment in model access, orchestration, and infrastructure rather than foundation-model construction 31. The industry thesis is becoming clear: value is migrating toward routing, APIs, deployment, governance, and workflow integration.
LangChain’s ecosystem illustrates this shift. In one support-platform deployment, a LangGraph router and fine-tuned Mistral 7B handled classification and 70% of first-draft replies 59. The competitive battleground therefore extends beyond Gemini itself. It includes the tools developers use to select, route, monitor, evaluate, and operationalize models. If the model layer becomes interchangeable, the orchestrator may command the customer relationship.
Enterprise AI financing is similarly targeted at specific operating problems. Dili raised $21.7 million in Series A financing 28,29,30, with Allianz and Rebel Fund participating 28,29. The financing is intended to automate compliance in infrastructure projects amid a construction and infrastructure boom 28. Freehand’s $75 million Series B was co-led by Battery Ventures and NewRoad Capital Partners 93. Gritt AI raised $32.4 million 80 to integrate robotic arms and autonomy into existing heavy equipment used in solar, data-center, and road construction 80. Ropedia raised $22 million 40,79 to scale real-world multimodal interaction data for embodied-AI robotics 40,79. Splash Robotics raised $4.2 million 80, while German robotics startup microagi raised €48 million in seed financing 27, reportedly the largest German startup seed round 27, and already works with Unitree and UBTECH 44.
Security and governance are developing into their own industrial layer. ThreatLocker completed a $190 million financing and Series F led by Elephant 77. Spur Intelligence raised $200 million from Insight Partners for cybersecurity, fraud prevention, traffic intelligence, and defensive AI 47. Hush Security raised $30 million 45,46 to accelerate ecosystem support and broaden corporate partnerships 45,46. JetStream Security launched a Verified MCP Governance Layer for enterprise agents 76. The company describes a high-growth market driven by enterprise-agent adoption, security concerns, and the transition from experimentation to production 76, claiming that its unified control plane addresses five AI trust gaps 76.
These developments create opportunities for Google Cloud security and compliance products. They also establish a limit on growth: customer trust in connected deployments could constrain AI adoption for platform vendors such as Elastic 75. In enterprise AI, governance is not a decorative feature. It is part of the infrastructure required to move a workload from demonstration to production.
Infrastructure Demand Is Strong, but Capacity and Financing Are Strategic Constraints
Chips, networking, storage, and power are becoming the new railroads
The funding cycle extends into the physical infrastructure required to operate AI. Lightmatter has raised $850 million cumulatively 85 and completed a $400 million Series D in October 2024 85. Lightelligence completed a Series C exceeding RMB 1.5 billion in September 2025 85, with reported participation from funds affiliated with China Mobile, Shanghai state-owned capital, China Reform Fund, and Tencent 85. Luminous Computing’s reported investors included Gigafund and Bill Gates 85. Israeli server-networking and storage-chip developer Xsight raised $300 million from Fidelity 62.
Verda Cloud announced financing and planned hardware investment 48, backed by the EU’s InvestEU program 48 and aligned with a broader European push into cloud, high-performance computing, and AI infrastructure 48. Abaxx Technologies expanded digital infrastructure alongside a capital raise 20, while Nocera has investments spanning AI infrastructure and energy storage or power 65. These are not peripheral suppliers. They are the equivalent of the rolling stock, transmission lines, and specialized machinery that determine the capacity of the industrial system.
Public policy is reinforcing the private capital cycle. Each first-lot AI Gigafactory project can receive up to €100 million in phase-one EU funding 64, while second-lot projects may qualify for up to an additional €800 million in phase two 64. EU initiatives such as Gaia-X and InvestAI reflect an effort to build regional infrastructure and investment capacity 26. South Korea is using public financing to support AI chips, cloud infrastructure, foundation models, and strategically important technology companies 41, with reinvestment of infrastructure savings into research and development described as a mechanism for improving Korean AI competitiveness 12. Texas has used tax incentives to position itself as an attractive location for AI infrastructure 35.
The implications for Alphabet are substantial. Location, cost, and availability of compute increasingly shape cloud competitiveness. Google’s TPU strategy and global data-center footprint can provide differentiation if customers value predictable capacity, lower cost, and regional deployment. A cloud provider that controls the accelerator, compiler, data center, and software interface possesses more bargaining power than one purchasing each component at market prices.
Google’s TPU advantage must become a commercial advantage
Recent performance data provide a concrete example. Google optimized inference for Mistral 3 Large, a mixture-of-experts model, on its Ironwood TPU platform, identified as TPU v7x 24,63. On Google Cloud, Mistral 3 Large inference performance improved 1.5 times 24, with up to 48% higher throughput 63, while benchmark accuracy remained neutral 63. The claims were published July 31 and the core optimization and performance improvement were corroborated by two sources.
This matters because it shows how Google Cloud can monetize infrastructure even when another company supplies the model. The TPU does not need to win every model contest if it can run a broad range of workloads more efficiently. The master resource is not merely model ownership; it is productive compute delivered at an attractive cost per useful inference.
The financial backdrop, however, is contested. S&P Global Ratings reported a 25% increase in investment-grade billed issuance, driven primarily by large AI-infrastructure financings and strong M&A issuance 21. One claim argues that the expansion is backed by massive free cash flow rather than speculative debt 34. Other evidence points to debt financing, off-balance-sheet structures, and leverage. Mistral’s $830 million debt financing 1,67 and Meta’s use of an off-balance-sheet special-purpose vehicle for an AI bond deal 81 demonstrate why financing structures require scrutiny. The claim that infrastructure fundraising is accelerating 23 is directionally consistent with the capital data, but the durability of that pace remains uncertain.
Open Models and Routing Increase Pressure on Alphabet
Model leadership is no longer sufficient for platform leadership
Mistral’s open-weight strategy 19,67,78 and reports that DeepSeek uses a training approach comparable to Google’s Gemma or Mistral 18 point to a more fragmented model market. DeepSeek is reportedly planning imminent fundraising 38, while Solar Open 2 is claimed to outperform DeepSeek V4 Flash and Mistral Medium 3.5 on agentic benchmarks 15. The Solar Open 2 result is supported by only two sources and should not be treated as conclusive evidence of durable superiority. It does, however, demonstrate how quickly model rankings and customer preferences can change.
The strategic implication is that proprietary model leadership may not automatically translate into economic leadership. Open-weight models reduce switching costs, enable local deployment, and allow customers to customize systems for regulated or sensitive environments. Mistral’s positioning around European data centers and models 78, local deployment, and data residency 78 may appeal to customers reluctant to rely exclusively on US-based providers.
Microsoft’s agreement with Mistral expands distribution 39, while Mistral is pursuing infrastructure support and distribution partnerships with major technology companies 39 and depends on those major partners 39. The Microsoft-Mistral relationship is framed as a cross-border technology and infrastructure response to strategic competition between Europe and the United States 78, rather than simply a model-licensing arrangement. That is an important distinction: distribution is becoming a strategic asset in its own right.
Alphabet should respond through platform integration rather than model exclusivity. Google can monetize third-party workloads through Google Cloud, use TPU efficiency to lower inference costs, and embed Gemini across Search, Workspace, Android, and other products. Yet customers may retain flexibility across providers and keep the orchestration layer themselves. The reported Stripe-OpenRouter interest 31 is especially relevant because it suggests that strategic value may accrue to the company aggregating models and managing access, not solely to the company training the largest model.
Distribution, orchestration, and neutrality may command the customer relationship
The emerging two-sided market has model suppliers on one side and enterprises and developers on the other. Routing platforms, cloud marketplaces, developer frameworks, and governance systems sit between them. LangChain’s funding 19, OpenRouter’s reported strategic value 31, Mistral’s Microsoft distribution agreement 39, and governance platforms such as JetStream 76 all point in the same direction.
Google Cloud’s opportunity is to become the default integrated platform while remaining open enough to host competing models. Its risk is that customers view compute as interchangeable and maintain their own model-selection layer, compressing differentiation and margins. The contest is no longer simply “Which model is best?” It is also “Who owns the API surface, deployment controls, observability, and enterprise workflow?”
Commercialization Is Advancing, but Durable Revenue Remains Uneven
Several claims indicate movement from pilots into production. Nebius is focused on helping enterprises migrate AI applications from pilot projects to production deployment 84. Axon management stated that every customer pitched on its AI product purchased it 72, and one unnamed company achieved multi-million-dollar enterprise wins 87 through Maestro, a platform for orchestrating complex AI interactions 87. Snowflake’s AI Data Cloud is identified as a potential growth catalyst 89. Saudi Arabia’s AI-native startup ecosystem benefits from strong national-policy support 82, and 48% of Saudi AI-native startups reportedly generate more than $400,000 of revenue per employee 82.
These signals are encouraging but heterogeneous. They include management commentary and single-source ecosystem estimates; they do not establish broad-based profitability. Pony AI faces commercialization risks as it moves from pilots to large-scale fleets, revenue generation, public adoption, and profitability 83, even as ARK Invest purchased more than 57,000 Pony AI shares 83 and Pony AI, WeRide, and Momenta entered the European market 71. Aurora Innovation targets an $80 million annualized revenue exit run-rate 90 but raised $215 million through an at-the-market issuance of 30 million shares 90, including $63 million used for cash bonuses and employee tax liabilities 90. The combination reveals both the opportunity in autonomous systems and their continuing dependence on external capital.
The same distinction between technical progress and commercial conversion applies to Alphabet. Investors should focus on paid workloads, retention, inference margins, and the proportion of customers moving from experimentation to production. A large installed base and strong benchmarks are not enough if pricing declines or if customers distribute workloads across several models and providers.
The strongest commercial opportunity appears where infrastructure improvements attach to specific enterprise workflows. Mistral’s applications in regulated industries 78, its OCR and workflow products 78, and Google Cloud’s inference efficiency gains 24,63 all point toward this model. Compute becomes more valuable when it is tied to compliance, document processing, security, construction, or another business process with a measurable return.
Capital-Market Stress Is the Necessary Counterweight
The infrastructure narrative must be tested against leverage and positioning risk. An AI-focused hedge fund associated with Leopold Aschenbrenner was reportedly valued at $45 billion in assets before losing most of its value within days 22. A separate claim says its assets fell to approximately $10 billion after liquidations to meet margin calls 62, while another indicates that Aschenbrenner used leverage to hold long positions in AI-infrastructure companies, including Micron 92. Other AI-fund selling could persist beyond the reported liquidation 91, and a leveraged fund was forced to liquidate after concentrated AI trades moved against it 73. The fund had attracted attention from Silicon Valley investors and Wall Street 94.
Capital also reportedly rotated from AI infrastructure into crypto stocks 32. The cluster describes a synchronized AI-financing and infrastructure unwind as a scenario in which private investment valuations would be marked down 69. These claims are largely single-source and scenario-oriented, so they should be weighted below the multi-source financing evidence. They nonetheless identify a central risk: even if end-user demand remains sound, a reversal in financing conditions could reduce private valuations, slow data-center commitments, pressure suppliers, and alter the economics of cloud expansion.
Alphabet is better insulated than early-stage companies because of its scale, cash generation, and diversified businesses. It is not immune. Large infrastructure commitments can produce weak returns if utilization lags, while aggressive competitor spending can force Google to invest defensively. One claim argues that Meta cannot benefit from AI infrastructure in the same way as hyperscalers 17; this is an isolated observation rather than a broad consensus, but it underscores the potential advantage of Alphabet’s integrated cloud and infrastructure model. Conversely, uncertainty over whether AI-related orders can continue to be fulfilled affects suppliers such as Marvell 88, demonstrating that supply constraints—not just demand—can shape returns across the system.
Capital Is Also Moving Across Adjacent Markets
Not every financing claim in the cluster is a direct AI comparable. Tenor raised $2.5 million to develop a blockchain-based lending platform 25. World raised $52.5 million through a token sale 39,66, following a prior $115 million round backed by Andreessen Horowitz, Khosla Ventures, and other prominent investors 66. The sale is framed as support for a global iris-scanned identity network 66 and as evidence of demand for identity infrastructure capable of authenticating humans amid AI-generated media 37,66.
Blockchain financing also includes Psalion’s $50 million Singapore VCC fund for seed-stage projects 55,57, focused on blockchain infrastructure, trade finance, real-world assets, and stablecoins 51. Repeated claims confirm that it targets pre-seed and seed companies 51,55. Beezie completed a $4 million round 55, while Birdai Labs raised a $4 million seed round led by Castle Island Ventures 54,56 to improve onchain trade execution 54,56. Bullet claims $13.5 million in funding, although the figure is unverified and its investors are unspecified 52,53. Reental raised €1.4 million from Akka and Monte Bianco Naif 50, while Tori raised $50 million in the seven days before launch 49.
Strategy generated $135 million by selling 2,225 bitcoin 58 and reportedly raised $17 billion in less than seven months 58, illustrating its dependence on continuous capital raising 58. Its digital-asset value rose from $49.7 billion to $54.8 billion between June 30 and July 27 58. An Indian fintech company that acquired Fisdom conducted an IPO 61, increased non-current investments 398.2% to ₹18,995.04 million through IPO-proceeds deployment and treasury operations 61, reduced current investments 51.6% to ₹7,380.93 million 61, and reported investing cash flow of negative ₹13,507.08 million versus positive ₹1,396.77 million 61.
These examples are not direct Alphabet comparables. They demonstrate, however, how quickly capital can migrate among AI, crypto, public equities, treasury assets, and adjacent infrastructure. Headline funding totals must therefore be assigned carefully. The relevant question is not whether capital is abundant in the abstract, but whether it is being committed to durable productive assets that Alphabet can monetize.
Defense, Autonomy, and Industrial AI Broaden the Market
AI investment is increasingly tied to defense and industrial systems. Mobileye acquired Mentee Robotics 6,79, while Ondas acquired Mistral and World View 86. Ondas’s Mistral business holds a position on a nearly $1 billion US Army IDIQ munitions program 86, exposing the company to government contracting rules and defense procurement processes 86. The acquisitions create integration risk 86. Leonardo DRS has a funded backlog reported at $4.7 billion 16, while a later claim places its Q2 2026 funded backlog at a record $5.1 billion, up 17% 73. DRS’s pending acquisition of Raft is intended to accelerate multi-domain AI, data fusion, and mission software 73. FTAI Aviation reported Q2 2026 MRE contract revenue of $182.8 million 74.
AI is also entering energy and advanced science. Commonwealth Fusion Systems raised another $1 billion, bringing total funding to $4 billion 62. TAE Technologies raised more than $150 million in its latest financing 68. GPUS began monetizing approximately 100 bitcoin to fund an AI-infrastructure project 43. Atoms secured a $1.7 billion investment led by Andreessen Horowitz, with participation from Bain Capital and Fifth Wall 80. FuriosaAI is targeting an IPO in the second half of the year 70. These claims are mostly single-source and should not be treated as equivalent in evidentiary strength. Collectively, however, they show capital flowing into the intersection of compute, autonomy, sensing, energy, defense, and scientific computing.
For Alphabet, the addressable cloud and AI-services market is consequently broader than conventional software. Defense, industrial automation, robotics, energy optimization, and scientific computing can generate high-value workloads. They also involve longer sales cycles, regulatory requirements, and difficult deployment environments. In these sectors, secure, compliant, scalable infrastructure may matter more than consumer-facing model novelty.
Funding Breadth Masks Wide Dispersion in Company Quality
The ecosystem includes companies with materially different funding histories and commercial positions. NavVis has raised $68.2 million 19, Noodle.AI $72 million 19, Harness $614 million 19, Hypatos $12 million 19, Zesty.ai $46 million 19, DataProphet $16 million 19, LogiNext $49.6 million 19, Aquify $36.8 million 19, DataRobot is described as the most funded AutoML company 19, MEGVII as a Chinese AI-engine developer 19, Mobvoi has raised $260 million 19, dbt Labs $416 million 19, ElevenLabs $282 million 19, Tractable $185 million 19, Nanox AI $74.1 million 19, Heuritech €5.2 million 19, and UiPath $2 billion 19.
Runway ML operates across generative-video models, real-time applications, APIs, enterprise products, and creative tools 10,11, with capabilities spanning applied research, model distillation, adversarial post-training, product engineering, infrastructure operations, enterprise management, and evaluation 11. Other early-stage companies include Neo, which raised $100 million through a $75 million Series A and a previously undisclosed $25 million seed round 8; Lightelligence’s large Series C 85; Harmony’s $34 million, Foundational Industries’ $25 million, and Terminal’s $20 million raises 77; Dymium, described as a secure AI-infrastructure company 9; and NEUROVATIC, which markets Trust Augmentation Infrastructure for AI, robotics, autonomous systems, and critical software 33. Space-Eyes has approximately $1 million in current annual revenue 60.
These companies are useful for mapping the industry but have limited direct valuation significance for Alphabet. Their dispersion is nevertheless informative. The AI economy is financing not only frontier laboratories, but also data infrastructure, observability, governance, enterprise automation, content generation, robotics, and security. Alphabet’s strongest structural position is likely in the shared infrastructure and distribution layers, while individual application categories remain competitive and vulnerable to commoditization.
Implications for Alphabet
1. Treat infrastructure as the primary strategic moat
The evidence supports a durable secular expansion in AI infrastructure. Private funding, public debt issuance, national subsidies, chip investment, cloud financing, and data-center construction all point toward continued ecosystem growth 21,23,64. Google Cloud’s 1.5x Mistral inference improvement and up to 48% throughput gain 24,63, without sacrificing benchmark accuracy 63, provide a tangible basis for monetizing TPU and cloud infrastructure independently of model ownership.
Alphabet should therefore measure its AI position by workload capture, TPU utilization, inference cost, and enterprise production revenue—not simply by benchmark rankings. If the company controls the accelerator, compiler, cloud, data services, and distribution channel, it can participate in the expansion of AI even when customers choose open-weight or third-party models.
2. Build an open but controlled platform
Mistral’s open-weight approach 19,67,78, along with DeepSeek and Solar Open competition 15,18, may pressure Gemini pricing and reduce the scarcity value of frontier models. Yet open models can also increase aggregate usage of compute, storage, networking, security, and data services. Alphabet’s most robust strategy is to maximize workload capture rather than insist that every workload use a Google-owned model.
That requires an integrated platform with sufficient neutrality to host competing models and sufficient differentiation to retain the customer. Google Cloud should make model choice, routing, governance, and regional deployment easy while using TPU economics and enterprise integration to preserve margin. In industrial terms, Alphabet should own the railroad even when it does not own every factory that ships goods across it.
3. Convert capital intensity into measured returns
Mistral’s large capital requirements 78, dependence on Nvidia and AMD 78, reported debt financing 1,67, and possible Samsung investment to secure memory access 42,67 illustrate the difficulty smaller laboratories face in scaling independently. Alphabet’s balance sheet, internal infrastructure, TPU development, and global distribution are structural advantages.
They are not a license for undisciplined spending. The magnitude of competitor financing and the possibility of leveraged unwinds 22,69,73,91 mean investors should evaluate returns on AI capital rather than capital deployed. A funding slowdown could damage weaker suppliers and improve Alphabet’s relative position. A sustained boom could intensify competitive spending and postpone profitability.
4. Focus enterprise expansion where trust and workflow create switching costs
Enterprise adoption is moving toward regulated and operationally consequential applications, including compliance 28,29,30, defense 78,86, robotics 27,40,79, security 47,77, industrial construction 80, and data-residency-sensitive deployments 78. These markets favor vendors that provide reliability, governance, regional infrastructure, and long-term support.
That is favorable for Google Cloud, but only if Alphabet converts technical capability into contracted production workloads and preserves customer trust. The risk that trust could limit connected-AI adoption 75, together with commercialization difficulties visible in autonomous driving 83,90, demonstrates that deployment economics remain unproven in many segments.
Conclusion
The AI capital cycle resembles the great infrastructure expansions of earlier industrial eras. Capital is laying down the equivalent of rail lines, building foundries for computation, and financing merchants and fabricators that will use the resulting capacity. Mistral’s reported €3 billion Series D at a €20 billion valuation is a prominent signal, but it remains unconfirmed and subject to execution risk 3,5,39,42,67,78. The broader and more durable signal is the expansion of investment across models, cloud, chips, orchestration, governance, robotics, defense, and enterprise applications.
For Alphabet, the opportunity is substantial. The company can benefit from AI demand even when customers use open-weight or third-party models, provided Google Cloud and the TPU ecosystem capture the underlying workloads. The central risks are equally clear: capital intensity, leverage, valuation excess, supply constraints, model commoditization, and weak conversion from pilots to production.
The most robust evidence points to accelerating AI infrastructure and software investment. The most speculative evidence concerns private valuations, unconfirmed financings, benchmark leadership, and the durability of capital flows. Investors should therefore monitor Google Cloud AI revenue, TPU utilization and pricing, third-party model consumption, enterprise production conversion, infrastructure returns, and competitor financing—not headline funding announcements alone. The decisive advantage will belong not to the company that raises the most capital, but to the company that converts capital into the lowest-cost, most deeply distributed, and most trusted means of computation.